Result:
found more than 109 distributions - search limited to the first 2001 files matching your query
( run in 0.899 )
AI-NeuralNet-Kohonen-Visual
view release on metacpan or search on metacpan
lib/AI/NeuralNet/Kohonen/Visual.pm view on Meta::CPAN
package AI::NeuralNet::Kohonen::Visual;
use vars qw/$VERSION/;
$VERSION = 0.3; # 05 May 2006 pod and packaging
=head1 NAME
AI::NeuralNet::Kohonen::Visual - Tk-based Visualisation
=head1 SYNOPSIS
Test the test file in this distribution, or:
package YourClass;
use base "AI::NeuralNet::Kohonen::Visual";
sub get_colour_for { my ($self,$x,$y) = (shift,shift,shift);
# From here you return a TK colour name.
# Get it as you please; for example, values of a 3D map:
return sprintf("#%02x%02x%02x",
(int (255 * $self->{map}->[$x]->[$y]->{weight}->[0])),
(int (255 * $self->{map}->[$x]->[$y]->{weight}->[1])),
(int (255 * $self->{map}->[$x]->[$y]->{weight}->[2])),
);
}
exit;
1;
And then:
use YourClass;
my $net = AI::NeuralNet::Kohonen::Visual->new(
display => 'hex',
map_dim => 39,
epochs => 19,
neighbour_factor => 2,
targeting => 1,
table => "3
1 0 0 red
0 1 0 yellow
0 0 1 blue
0 1 1 cyan
1 1 0 yellow
1 .5 0 orange
1 .5 1 pink",
);
$net->train;
$net->plot_map;
$net->main_loop;
exit;
=head1 DESCRIPTION
Provides TK-based visualisation routines for C<AI::NueralNet::Kohonen>.
Replaces the earlier C<AI::NeuralNet::Kohonen::Demo::RGB>.
This is a sub-class of C<AI::NeuralNet::Kohonen>
that impliments extra methods to make use of TK.
This moudle is itself intended to be sub-classed by you,
where you provide a version of the method C<get_colour_for>:
see L<METHOD get_colour_for> and L<SYNOPSIS> for details.
=head1 CONSTRUCTOR (new)
The following paramter fields are added to the base module's fields:
=over 4
=item display
Set to C<hex> for display as a unified distance matrix, rather than
as the default plain grid;
=item display_scale
Set with a factor to effect the size of the display.
=item show_bmu
Show the current BMU during training.
=item show_training
Display updates during training.
=item label_bmu
=item label_all
Displays labels...
=item MainLoop
Calls TK's C<MainLoop> at the end of training.
=item missing_colour
When selecting a colour using L<METHOD get_colour_for>,
every node weight holding the value of C<missing_mask>
will be given the value of this paramter. If this paramter
is not defined, the default is 0.
=back
=cut
use strict;
use warnings;
use Carp qw/cluck carp confess croak/;
use base "AI::NeuralNet::Kohonen";
use Tk;
use Tk::Canvas;
use Tk::Label;
use Tk qw/DoOneEvent DONT_WAIT/;
=head1 METHOD train
Over-rides the base class to provide TK displays of the map.
=cut
sub train { my ($self,$epochs) = (shift,shift);
$epochs = $self->{epochs} unless defined $epochs;
$self->{display_scale} = 10 if not defined $self->{display_scale};
&{$self->{train_start}} if exists $self->{train_start};
$self->prepare_display if not defined $self->{_mw} or ref $self->{_mw} ne 'MainWindow';
# Replaces Tk's MainLoop
for (0..$self->{epochs}) {
if ($self->{_quit_flag}) {
$self->{_mw}->destroy;
$self->{_mw} = undef;
return;
}
$self->{t}++; # Measure epoch
&{$self->{epoch_start}} if exists $self->{epoch_start};
for (0..$#{$self->{input}}){
my $target = $self->_select_target;
my $bmu = $self->find_bmu($target);
$self->_adjust_neighbours_of($bmu,$target);
if (exists $self->{show_training}){
if ($self->{show_bmu}){
$self->plot_map(bmu_x=>$bmu->[1],bmu_y=>$bmu->[2]);
} else {
$self->plot_map;
}
$self->{_label_txt} = sprintf("Epoch: %04d",$self->{t})." "
. "Learning: $self->{l} "
. sprintf("BMU: %02d,%02d",$bmu->[1],$bmu->[2])." "
.( exists $target->{class}? "Target: [$target->{class}] " : "")
;
$self->{_canvas}->update;
$self->{_label}->update;
DoOneEvent(DONT_WAIT); # be kind and process XEvents if they arise
}
}
$self->_decay_learning_rate;
&{$self->{epoch_end}} if exists $self->{epoch_end};
}
$self->{_label_txt} = "Did $self->{t} epochs: ";
$self->{_label_txt} .= "now smoothed." if $self->{smoothing};
$_->smooth if $self->{smooth};
$self->plot_map if $self->{MainLoop};
&{$self->{train_end}} if exists $self->{train_end};
MainLoop if $self->{MainLoop};
return 1;
}
=head1 METHOD get_colour_for
This method is intended to be sub-classed.
Currently it only operates on the first three elements
of a weight vector, turning them into RGB values.
It returns the a TK colour for a node at position C<x>,C<y> in the
C<map> paramter.
Accepts: C<x> and C<y> co-ordinates in the map.
=cut
sub get_colour_for { my ($self,$x,$y) = (shift,shift,shift);
my $_0 = $self->{map}->[$x]->[$y]->{weight}->[0];
$_0 = $self->{missing_colour} || 0 if $_0 eq $self->{missing_mask};
my $_1 = $self->{map}->[$x]->[$y]->{weight}->[1];
$_1 = $self->{missing_colour} || 0 if $_1 eq $self->{missing_mask};
my $_2 = $self->{map}->[$x]->[$y]->{weight}->[2];
$_2 = $self->{missing_colour} || 0 if $_2 eq $self->{missing_mask};
return sprintf("#%02x%02x%02x",
(int (255 * $_0)),
(int (255 * $_1)),
(int (255 * $_2)),
);
}
=head1 METHOD prepare_display
Depracated: see L<METHOD create_empty_map>.
=cut
sub prepare_display {
return $_[0]->create_empty_map;
}
=head1 METHOD create_empty_map
Sets up a TK C<MainWindow> and C<Canvas> to
act as an empty map.
=cut
sub create_empty_map { my $self = shift;
my ($w,$h);
if ($self->{display} and $self->{display} eq 'hex'){
$w = ($self->{map_dim_x}+1) * ($self->{display_scale}+2);
$h = ($self->{map_dim_y}+1) * ($self->{display_scale}+2);
} else {
$w = ($self->{map_dim_x}+1) * ($self->{display_scale});
$h = ($self->{map_dim_y}+1) * ($self->{display_scale});
}
$self->{_mw} = MainWindow->new(
-width => $w + 20,
-height => $h + 20,
);
$self->{_mw}->fontCreate(qw/TAG -family verdana -size 8 -weight bold/);
$self->{_mw}->resizable( 0, 0);
$self->{_quit_flag} = 0;
$self->{_mw}->protocol('WM_DELETE_WINDOW' => sub {$self->{_quit_flag}=1});
$self->{_canvas} = $self->{_mw}->Canvas(
-width => $w,
-height => $h,
-relief => 'raised',
-border => 2,
);
$self->{_canvas}->pack(-side=>'top');
$self->{_label} = $self->{_mw}->Button(
-command => sub { $self->{_mw}->destroy;$self->{_mw} = undef; },
-relief => 'groove',
-text => ' ',
-wraplength => $w,
-textvariable => \$self->{_label_txt}
);
$self->{_label}->pack(-side=>'top');
return 1;
}
=head1 METHOD plot_map
Plots the map on the existing canvas. Arguments are supplied
in a hash with the following keys as options:
The values of C<bmu_x> and C<bmu_y> represent The I<x> and I<y>
co-ordinates of unit to highlight using the value in the
C<hicol> to highlight it with colour. If no C<hicolo> is provided,
it default to red.
When called, this method also sets the object field flag C<plotted>:
currently, this prevents C<main_loop> from calling this routine.
See also L<METHOD get_colour_for>.
=cut
sub plot_map { my ($self,$args) = (shift,{@_});
$self->{plotted} = 1;
# MW may have been destroyed
$self->prepare_display if not defined $self->{_mw};
my $yo = 5+($self->{display_scale}/2);
for my $x (0..$self->{map_dim_x}){
for my $y (0..$self->{map_dim_y}){
my $colour;
if ($args->{bmu_x} and $args->{bmu_x}==$x and $args->{bmu_y}==$y){
$colour = $args->{hicol} || 'red';
} else {
$colour = $self->get_colour_for ($x,$y);
}
if ($self->{display} and $self->{display} eq 'hex'){
my $xo = 5+($y % 2) * ($self->{display_scale}/2);
$self->{_canvas}->create(
polygon => [
$xo + (($x)*$self->{display_scale} ),
$yo + (($y)*$self->{display_scale} ),
# polygon only:
$xo + (($x)*($self->{display_scale})+($self->{display_scale}/2) ),
$yo + (($y)*($self->{display_scale})-($self->{display_scale}/2) ),
#
$xo + (($x)*($self->{display_scale})+$self->{display_scale} ),
$yo + (($y)*$self->{display_scale} ),
$xo + (($x)*($self->{display_scale})+$self->{display_scale} ),
$yo + (($y)*($self->{display_scale})+($self->{display_scale}/2) ),
# Polygon only:
$xo + (($x)*($self->{display_scale})+($self->{display_scale}/2) ),
$yo + (($y)*($self->{display_scale})+($self->{display_scale}) ),
#
$xo + (($x)*$self->{display_scale} ),
$yo + (($y)*($self->{display_scale})+($self->{display_scale}/2) ),
],
-outline => "black",
-fill => $colour,
);
}
else {
$self->{_canvas}->create(
rectangle => [
$x*$self->{display_scale} +1,
$y*$self->{display_scale} +1,
$x*($self->{display_scale})+$self->{display_scale} +1,
$y*($self->{display_scale})+$self->{display_scale} +1
],
-outline => "black",
-fill => $colour,
);
}
# Label
if ($self->{label_all}){
my $txt;
unless ( $txt = $self->{map}->[$x]->[$y]->{class}){
$txt = "";
}
$self->label_map($x,$y,"+$txt");
}
}
}
if ($self->{label_bmu}){
my $txt;
unless ( $txt = $self->{map}->[$args->{bmu_x}]->[$args->{bmu_y}]->{class}){
$txt = "";
}
$self->label_map(
$args->{bmu_x}, $args->{bmu_y}, "+$txt"
);
}
$self->{_canvas}->update;
$self->{_label}->update;
return 1;
}
=head1 METHOD label_map
Put a text label on the map for the node at the I<x,y> co-ordinates
supplied in the first two paramters, using the text supplied in the
third.
Very naive: no attempt to check the text will appear on the map.
=cut
sub label_map { my ($self,$x,$y,$t) = (shift,shift,shift,shift);
$self->{_canvas}->createText(
$x*$self->{display_scale}+($self->{display_scale}),
$y*$self->{display_scale}+($self->{display_scale}),
-text => $t,
-anchor => 'w',
-fill => 'white',
-font => 'TAG',
);
}
=head1 METHOD main_loop
Calls TK's C<MainLoop> to keep a window open until the user closes it.
=cut
sub main_loop { my $self = shift;
$self->plot_map unless $self->{plotted};
MainLoop;
}
1;
__END__
=head1 SEE ALSO
See
L<AI::NeuralNet::Kohonen>;
L<AI::NeuralNet::Kohonen::Node>;
=head1 AUTHOR AND COYRIGHT
This implimentation Copyright (C) Lee Goddard, 2003.
All Rights Reserved.
Available under the same terms as Perl itself.
AI-NeuralNet-Mesh
view release on metacpan or search on metacpan
#!/usr/bin/perl
# Copyright (c) 2000 Josiah Bryan USA
#
# See AUTHOR section in pod text below for usage and distribution rights.
#
BEGIN {
$AI::NeuralNet::Mesh::VERSION = "0.44";
$AI::NeuralNet::Mesh::ID =
'$Id: AI::NeuralNet::Mesh.pm, v'.$AI::NeuralNet::Mesh::VERSION.' 2000/15/09 03:29:08 josiah Exp $';
}
package AI::NeuralNet::Mesh;
require Exporter;
@ISA = qw(Exporter);
@EXPORT = qw(range intr pdiff);
%EXPORT_TAGS = (
'default' => [ qw ( range intr pdiff )],
'all' => [ qw ( p low high ramp and_gate or_gate range intr pdiff ) ],
'p' => [ qw ( p low high intr pdiff ) ],
'acts' => [ qw ( ramp and_gate or_gate range ) ],
);
@EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} }, qw( p low high ramp and_gate or_gate ) );
use strict;
use Benchmark;
# See POD for usage of this variable.
$AI::NeuralNet::Mesh::Connector = '_c';
# Debugging subs
$AI::NeuralNet::Mesh::DEBUG = 0;
sub whowasi { (caller(1))[3] . '()' }
sub debug { shift; $AI::NeuralNet::Mesh::DEBUG = shift || 0; }
sub d { shift if(substr($_[0],0,4) eq 'AI::'); my ($a,$b,$c)=(shift,shift,$AI::NeuralNet::Mesh::DEBUG); print $a if($c == $b); return $c }
sub verbose {debug @_};
sub verbosity {debug @_};
sub v {debug @_};
# Return version of ::ID string passed or current version of this
# module if no string is passed. Used in load() to detect file versions.
sub version {
shift if(substr($_[0],0,4) eq 'AI::');
substr((split(/\s/,(shift || $AI::NeuralNet::Mesh::ID)))[2],1);
}
# Rounds a floating-point to an integer with int() and sprintf()
sub intr {
shift if(substr($_[0],0,4) eq 'AI::');
try { return int(sprintf("%.0f",shift)) }
catch { return 0 }
}
# Package constructor
sub new {
no strict 'refs';
my $type = shift;
my $self = {};
my $layers = shift;
my $nodes = shift;
my $outputs = shift || $nodes;
my $inputs = shift || $nodes;
bless $self, $type;
# If $layers is a string, then it will be numerically equal to 0, so
# try to load it as a network file.
if($layers == 0) {
# We use a "1" flag as the second argument to indicate that we
# want load() to call the new constructor to make a network the
# same size as in the file and return a refrence to the network,
# instead of just creating the network from pre-exisiting refrence
return $self->load($layers,1);
}
# Looks like we got ourselves a layer specs array
if(ref($layers) eq "ARRAY") {
if(ref($layers->[0]) eq "HASH") {
$self->{total_nodes} = 0;
$self->{inputs} = $layers->[0]->{nodes};
$self->{nodes} = $layers->[0]->{nodes};
$self->{outputs} = $layers->[$#{$layers}]->{nodes};
$self->{total_layers} = $#{$layers};
for (0..$#{$layers}){$self->{layers}->[$_] = $layers->[$_]->{nodes}}
for (0..$self->{total_layers}){$self->{total_nodes}+=$self->{layers}->[$_]}
} else {
$self->{inputs} = $layers->[0];
$self->{nodes} = $layers->[0];
$self->{outputs} = $layers->[$#{$layers}];
$self->{layers} = $layers;
$self->{total_layers} = $#{$self->{layers}};
$self->{total_nodes} = 0;
for (0..$self->{total_layers}) {
$self->{total_nodes}+=$self->{layers}->[$_];
}
}
} else {
$self->{total_nodes} = $layers * $nodes + $outputs;
$self->{total_layers} = $layers;
$self->{nodes} = $nodes;
$self->{inputs} = $inputs;
$self->{outputs} = $outputs;
}
# Initalize misc. variables
$self->{col_width} = 5;
$self->{random} = 0;
$self->{const} = 0.0001;
$self->{connector} = $AI::NeuralNet::Mesh::Connector;
# Build mesh
$self->_init();
# Initalize activation, thresholds, etc, if provided
if(ref($layers->[0]) eq "HASH") {
for (0..$self->{total_layers}) {
$self->activation($_,$layers->[$_]->{activation});
$self->threshold($_,$layers->[$_]->{threshold});
$self->mean($_,$layers->[$_]->{mean});
}
}
# Done!
return $self;
}
# Internal usage
# Connects one range of nodes to another range
sub _c {
my $self = shift;
my $r1a = shift;
my $r1b = shift;
my $r2a = shift;
my $r2b = shift;
my $m1 = shift || $self->{mesh};
my $m2 = shift || $m1;
for my $y ($r1a..$r1b-1) {
for my $z ($r2a..$r2b-1) {
$m1->[$y]->add_output_node($m2->[$z]);
}
}
}
# Internal usage
# Creates the mesh of neurons
sub _init {
my $self = shift;
my $nodes = $self->{nodes};
my $outputs = $self->{outputs} || $nodes;
my $inputs = $self->{inputs} || $nodes;
my $layers = $self->{total_layers};
my $tmp = $self->{total_nodes} || ($layers * $nodes + $outputs);
my $layer_specs = $self->{layers};
my $connector = $self->{connector};
my ($x,$y,$z);
no strict 'refs';
# Just to be safe.
$self->{total_nodes} = $tmp;
# If they didn't give layer specifications, then we derive our own specs.
if(!(defined $self->{layers})) {
$layer_specs = [split(',',"$nodes," x $layers)];
$layer_specs->[$#{$layer_specs}+1]=$outputs;
$self->{layers} = $layer_specs;
}
# First create the individual nodes
for my $x (0..$tmp-1) {
$self->{mesh}->[$x] = AI::NeuralNet::Mesh::node->new($self);
}
# Get an instance of an output (data collector) node
$self->{output} = AI::NeuralNet::Mesh::output->new($self);
# Connect the output layer to the data collector
for $x (0..$outputs-1) {
$self->{mesh}->[$tmp-$outputs+$x]->add_output_node($self->{output});
}
# Now we use the _c() method to connect the layers together.
$y=0;
my $c = $connector.'($self,$y,$y+$z,$y+$z,$y+$z+$layer_specs->[$x+1])';
for $x (0..$layers-1) {
$z = $layer_specs->[$x];
d("layer $x size: $z (y:$y)\n,",1);
eval $c;
$y+=$z;
}
# Get an instance of our cap node.
$self->{input}->{cap} = AI::NeuralNet::Mesh::cap->new();
# Add a cap to the bottom of the mesh to stop it from trying
# to recursivly adjust_weight() where there are no more nodes.
for my $x (0..$inputs-1) {
$self->{input}->{IDs}->[$x] =
$self->{mesh}->[$x]->add_input_node($self->{input}->{cap});
}
}
# See POD for usage
sub extend {
my $self = shift;
my $layers = shift;
# Looks like we got ourselves a layer specs array
if(ref($layers) eq "ARRAY") {
if($self->{total_layers}!=$#{$layers}) {
$self->{error} = "extend(): Cannot add new layers. Create a new network to add layers.\n";
return undef;
}
if(ref($layers->[0]) eq "HASH") {
$self->{total_nodes} = 0;
$self->{inputs} = $layers->[0]->{nodes};
$self->{nodes} = $layers->[0]->{nodes};
$self->{outputs} = $layers->[$#{$layers}]->{nodes};
for (0..$#{$layers}){
$self->extend_layer($_,$layers->[$_]);
$self->{layers}->[$_] =$layers->[$_]->{nodes};
}
for (0..$self->{total_layers}){$self->{total_nodes}+=$self->{layers}->[$_]}
} else {
$self->{inputs} = $layers->[0];
$self->{nodes} = $layers->[0];
$self->{outputs} = $layers->[$#{$layers}];
$self->{total_nodes} = 0;
for (0..$self->{total_layers}){$self->extend_layer($_,$layers->[$_])}
$self->{layers} = $layers;
for (0..$self->{total_layers}){$self->{total_nodes}+= $self->{layers}->[$_]}
}
} else {
$self->{error} = "extend(): Invalid argument type.\n";
return undef;
}
return 1;
}
# See POD for usage
sub extend_layer {
my $self = shift;
my $layer = shift || 0;
my $specs = shift;
if(!$specs) {
$self->{error} = "extend_layer(): You must provide specs to extend layer $layer with.\n";
return undef;
}
if(ref($specs) eq "HASH") {
$self->activation($layer,$specs->{activation}) if($specs->{activation});
$self->threshold($layer,$specs->{threshold}) if($specs->{threshold});
$self->mean($layer,$specs->{mean}) if($specs->{mean});
return $self->add_nodes($layer,$specs->{nodes});
} else {
return $self->add_nodes($layer,$specs);
}
return 1;
}
# Pseudo-internal usage
sub add_nodes {
no strict 'refs';
my $self = shift;
my $layer = shift;
my $nodes = shift;
my $n = 0;
my $more = $nodes - $self->{layers}->[$layer] - 1;
d("Checking on extending layer $layer to $nodes nodes (check:$self->{layers}->[$layer]).\n",9);
return 1 if ($nodes == $self->{layers}->[$layer]);
if ($self->{layers}->[$layer]>$nodes) {
$self->{error} = "add_nodes(): I cannot remove nodes from the network with this version of my module. You must create a new network to remove nodes.\n";
return undef;
}
d("Extending layer $layer by $more.\n",9);
for (0..$more){$self->{mesh}->[$#{$self->{mesh}}+1]=AI::NeuralNet::Mesh::node->new($self)}
for(0..$layer-2){$n+=$self->{layers}->[$_]}
$self->_c($n,$n+$self->{layers}->[$layer-1],$#{$self->{mesh}}-$more+1,$#{$self->{mesh}});
$self->_c($#{$self->{mesh}}-$more+1,$#{$self->{mesh}},$n+$self->{layers}->[$layer],$n+$self->{layers}->[$layer]+$self->{layers}->[$layer+1]);
}
# See POD for usage
sub run {
my $self = shift;
my $inputs = shift;
my $const = $self->{const};
#my $start = new Benchmark;
$inputs = $self->crunch($inputs) if($inputs == 0);
no strict 'refs';
for my $x (0..$#{$inputs}) {
last if($x>$self->{inputs});
d("inputing $inputs->[$x] at index $x with ID $self->{input}->{IDs}->[$x].\n",1);
$self->{mesh}->[$x]->input($inputs->[$x]+$const,$self->{input}->{IDs}->[$x]);
}
if($#{$inputs}<$self->{inputs}-1) {
for my $x ($#{$inputs}+1..$self->{inputs}-1) {
d("inputing 1 at index $x with ID $self->{input}->{IDs}->[$x].\n",1);
$self->{mesh}->[$x]->input(1,$self->{input}->{IDs}->[$x]);
}
}
#$self->{benchmark} = timestr(timediff(new Benchmark, $start));
return $self->{output}->get_outputs();
}
# See POD for usage
sub run_uc {
$_[0]->uncrunch(run(@_));
}
# See POD for usage
sub learn {
my $self = shift;
my $inputs = shift; # input set
my $outputs = shift; # target outputs
my %args = @_; # get args into hash
my $inc = $args{inc} || 0.002; # learning gradient
my $max = $args{max} || 1024; # max iteterations
my $degrade = $args{degrade} || 0; # enable gradient degrading
my $error = ($args{error}>-1 && defined $args{error}) ? $args{error} : -1;
my $dinc = 0.0002; # amount to adjust gradient by
my $diff = 100; # error magin between results
my $start = new Benchmark;
$inputs = $self->crunch($inputs) if($inputs == 0);
$outputs = $self->crunch($outputs) if($outputs == 0);
my ($flag,$ldiff,$cdiff,$_mi,$loop,$y);
while(!$flag && ($max ? $loop<$max : 1)) {
my $b = new Benchmark;
my $got = $self->run($inputs);
$diff = pdiff($got,$outputs);
$flag = 1;
if(($error>-1 ? $diff<$error : 0) || !$diff) {
$flag=1;
last;
}
if($degrade) {
$inc -= ($dinc*$diff);
if($diff eq $ldiff) {
$cdiff++;
$inc += ($dinc*$diff)+($dinc*$cdiff*10);
} else {
$cdiff=0;
}
$ldiff = $diff;
}
for my $x (0..$self->{outputs}-1) {
my $a = $got->[$x];
my $b = $outputs->[$x];
d("got: $a, wanted: $b\n",2);
if ($a != $b) {
$flag = 0;
$y = $self->{total_nodes}-$self->{outputs}+$x;
$self->{mesh}->[$y]->adjust_weight(($a<$b?1:-1)*$inc,$b);
}
}
$loop++;
d("===============================Loop: [$loop]===================================\n",4);
d("Current Error: $diff\tCurrent Increment: $inc\n",4);
d("Benchmark: ".timestr(timediff(new Benchmark,$b))."\n",4);
d("============================Results, [$loop]===================================\n",4);
d("Actual: ",4);
join_cols($got,($self->{col_width})?$self->{col_width}:5) if(d()==4);
d("Target: ",4);
join_cols($outputs,($self->{col_width})?$self->{col_width}:5) if(d()==4);
d("\n",4);
d('.',12);
d('['.join(',',@{$got})."-".join(',',@{$outputs}).']',13);
}
my $str = "Learning took $loop loops and ".timestr(timediff(new Benchmark,$start))."\n";
d($str,3); $self->{benchmark} = "$loop loops and ".timestr(timediff(new Benchmark,$start))."\n";
return $str;
}
# See POD for usage
sub learn_set {
my $self = shift;
my $data = shift;
my %args = @_;
my $len = $#{$data}/2;
my $inc = $args{inc};
my $max = $args{max};
my $error = $args{error};
my $degrade = $args{degrade};
my $p = (defined $args{flag}) ?$args{flag} :1;
my $row = (defined $args{row}) ?$args{row}+1:1;
my $leave = (defined $args{leave})?$args{leave}:0;
for my $x (0..$len-$leave) {
d("Learning set $x...\n",4);
my $str = $self->learn( $data->[$x*2],
$data->[$x*2+1],
inc=>$inc,
max=>$max,
error=>$error,
degrade=>$degrade);
}
if ($p) {
return pdiff($data->[$row],$self->run($data->[$row-1]));
} else {
return $data->[$row]->[0]-$self->run($data->[$row-1])->[0];
}
}
# See POD for usage
sub run_set {
my $self = shift;
my $data = shift;
my $len = $#{$data}/2;
my (@results,$res);
for my $x (0..$len) {
$res = $self->run($data->[$x*2]);
for(0..$#{$res}){$results[$x]->[$_]=$res->[$_]}
d("Running set $x [$res->[0]]...\r",4);
}
return \@results;
}
#
# Loads a CSV-like dataset from disk
#
# Usage:
# my $set = $set->load_set($file, $column, $seperator);
#
# Returns a data set of the same format as required by the
# learn_set() method. $file is the disk file to load set from.
# $column an optional variable specifying the column in the
# data set to use as the class attribute. $class defaults to 0.
# $seperator is an optional variable specifying the seperator
# character between values. $seperator defaults to ',' (a single comma).
# NOTE: This does not handle quoted fields, or any other record
# seperator other than "\n".
#
sub load_set {
my $self = shift;
my $file = shift;
my $attr = shift || 0;
my $sep = shift || ',';
my $data = [];
open(FILE, $file);
my @lines = <FILE>;
close(FILE);
for my $x (0..$#lines) {
chomp($lines[$x]);
my @tmp = split /$sep/, $lines[$x];
my $c=0;
for(0..$#tmp){
$tmp[$_]=$self->crunch($tmp[$_])->[0] if($tmp[$_]=~/[AaBbCcDdEeFfGgHhIiJjKkLlMmNnOoPpQqRrSsTtUuVvWwXxYyZz]/);
if($_!=$attr){$data->[$x*2]->[$c]=$tmp[$c];$c++}
};
d("Loaded line $x, [@tmp] \r",4);
$data->[$x*2+1]=[$tmp[$attr]];
}
return $data;
}
# See POD for usage
sub get_outs {
my $self = shift;
my $data = shift;
my $len = $#{$data}/2;
my $outs = [];
for my $x (0..$len) {
$outs->[$x] = $data->[$x*2+1];
}
return $outs;
}
# Save entire network state to disk.
sub save {
my $self = shift;
my $file = shift;
no strict 'refs';
open(FILE,">$file");
print FILE "header=$AI::NeuralNet::Mesh::ID\n";
print FILE "total_layers=$self->{total_layers}\n";
print FILE "total_nodes=$self->{total_nodes}\n";
print FILE "nodes=$self->{nodes}\n";
print FILE "inputs=$self->{inputs}\n";
print FILE "outputs=$self->{outputs}\n";
print FILE "layers=",(($self->{layers})?join(',',@{$self->{layers}}):''),"\n";
print FILE "rand=$self->{random}\n";
print FILE "const=$self->{const}\n";
print FILE "cw=$self->{col_width}\n";
print FILE "crunch=$self->{_crunched}->{_length}\n";
print FILE "rA=$self->{rA}\n";
print FILE "rB=$self->{rB}\n";
print FILE "rS=$self->{rS}\n";
print FILE "rRef=",(($self->{rRef})?join(',',@{$self->{rRef}}):''),"\n";
for my $a (0..$self->{_crunched}->{_length}-1) {
print FILE "c$a=$self->{_crunched}->{list}->[$a]\n";
}
my $n = 0;
for my $x (0..$self->{total_layers}) {
for my $y (0..$self->{layers}->[$x]-1) {
my $w='';
for my $z (0..$self->{layers}->[$x-1]-1) {
$w.="$self->{mesh}->[$n]->{_inputs}->[$z]->{weight},";
}
print FILE "n$n=$w$self->{mesh}->[$n]->{activation},$self->{mesh}->[$n]->{threshold},$self->{mesh}->[$n]->{mean}\n";
$n++;
}
}
close(FILE);
if(!(-f $file)) {
$self->{error} = "Error writing to \"$file\".";
return undef;
}
return $self;
}
# Load entire network state from disk.
sub load {
my $self = shift;
my $file = shift;
my $load_flag = shift;
my @lines;
if(-f $file) {
open(FILE,"$file");
@lines=<FILE>;
close(FILE);
} else {
@lines=split /\n/, $file;
}
my %db;
for my $line (@lines) {
chomp($line);
my ($a,$b) = split /=/, $line;
$db{$a}=$b;
}
if(!$db{"header"}) {
$self->{error} = "Invalid format.";
return undef;
}
return $self->load_old($file) if($self->version($db{"header"})<0.21);
if($load_flag) {
undef $self;
$self = AI::NeuralNet::Mesh->new([split(',',$db{layers})]);
} else {
$self->{inputs} = $db{inputs};
$self->{nodes} = $db{nodes};
$self->{outputs} = $db{outputs};
$self->{layers} = [split(',',$db{layers})];
$self->{total_layers} = $db{total_layers};
$self->{total_nodes} = $db{total_nodes};
}
# Load variables
$self->{random} = $db{"rand"};
$self->{const} = $db{"const"};
$self->{col_width} = $db{"cw"};
$self->{rA} = $db{"rA"};
$self->{rB} = $db{"rB"};
$self->{rS} = $db{"rS"};
$self->{rRef} = [split /\,/, $db{"rRef"}];
$self->{_crunched}->{_length} = $db{"crunch"};
for my $a (0..$self->{_crunched}->{_length}-1) {
$self->{_crunched}->{list}->[$a] = $db{"c$a"};
}
$self->_init();
my $n = 0;
for my $x (0..$self->{total_layers}) {
for my $y (0..$self->{layers}->[$x]-1) {
my @l = split /\,/, $db{"n$n"};
for my $z (0..$self->{layers}->[$x-1]-1) {
$self->{mesh}->[$n]->{_inputs}->[$z]->{weight} = $l[$z];
}
my $z = $self->{layers}->[$x-1];
$self->{mesh}->[$n]->{activation} = $l[$z];
$self->{mesh}->[$n]->{threshold} = $l[$z+1];
$self->{mesh}->[$n]->{mean} = $l[$z+2];
$n++;
}
}
return $self;
}
# Load entire network state from disk.
sub load_old {
my $self = shift;
my $file = shift;
my $load_flag = shift;
if(!(-f $file)) {
$self->{error} = "File \"$file\" does not exist.";
return undef;
}
open(FILE,"$file");
my @lines=<FILE>;
close(FILE);
my %db;
for my $line (@lines) {
chomp($line);
my ($a,$b) = split /=/, $line;
$db{$a}=$b;
}
if(!$db{"header"}) {
$self->{error} = "Invalid format.";
return undef;
}
if($load_flag) {
undef $self;
# Create new network
$self = AI::NeuralNet::Mesh->new($db{"layers"},
$db{"nodes"},
$db{"outputs"});
} else {
$self->{total_layers} = $db{"layers"};
$self->{nodes} = $db{"nodes"};
$self->{outputs} = $db{"outputs"};
$self->{inputs} = $db{"nodes"};
#$self->{total_nodes} = $db{"total"};
}
# Load variables
$self->{random} = $db{"rand"};
$self->{const} = $db{"const"};
$self->{col_width} = $db{"cw"};
$self->{rA} = $db{"rA"};
$self->{rB} = $db{"rB"};
$self->{rS} = $db{"rS"};
$self->{rRef} = [split /\,/, $db{"rRef"}];
$self->{_crunched}->{_length} = $db{"crunch"};
for my $a (0..$self->{_crunched}->{_length}-1) {
$self->{_crunched}->{list}->[$a] = $db{"c$a"};
}
$self->_init();
my $nodes = $self->{nodes};
my $outputs = $self->{outputs};
my $tmp = $self->{total_nodes};
my $div = intr($nodes/$outputs);
# Load input and hidden
for my $a (0..$tmp-1) {
my @l = split /\,/, $db{"n$a"};
for my $b (0..$nodes-1) {
$self->{mesh}->[$a]->{_inputs}->[$b]->{weight} = $l[$b];
}
}
# Load output layer
for my $x (0..$outputs-1) {
my @l = split /\,/, $db{"n".($tmp+$x)};
for my $y (0..$div-1) {
$self->{mesh}->[$tmp+$x]->{_inputs}->[$y]->{weight} = $l[$y];
}
}
return $self;
}
# Dumps the complete weight matrix of the network to STDIO
sub show {
my $self = shift;
my $n = 0;
no strict 'refs';
for my $x (0..$self->{total_layers}) {
for my $y (0..$self->{layers}->[$x]-1) {
for my $z (0..$self->{layers}->[$x-1]-1) {
print "$self->{mesh}->[$n]->{_inputs}->[$z]->{weight},";
}
$n++;
}
print "\n";
}
}
# Set the activation type of a specific layer.
# usage: $net->activation($layer,$type);
# $type can be: "linear", "sigmoid", "sigmoid_2".
# You can use "sigmoid_1" as a synonym to "sigmoid".
# Type can also be a CODE ref, ( ref($type) eq "CODE" ).
# If $type is a CODE ref, then the function is called in this form:
# $output = &$type($sum_of_inputs,$self);
# The code ref then has access to all the data in that node (thru the
# blessed refrence $self) and is expected to return the value to be used
# as the output for that node. The sum of all the inputs to that node
# is already summed and passed as the first argument.
sub activation {
my $self = shift;
my $layer = shift || 0;
my $value = shift || 'linear';
my $n = 0;
no strict 'refs';
for(0..$layer-1){$n+=$self->{layers}->[$_]}
for($n..$n+$self->{layers}->[$layer]-1) {
$self->{mesh}->[$_]->{activation} = $value;
}
}
# Applies an activation type to a specific node
sub node_activation {
my $self = shift;
my $layer = shift || 0;
my $node = shift || 0;
my $value = shift || 'linear';
my $n = 0;
no strict 'refs';
for(0..$layer-1){$n+=$self->{layers}->[$_]}
$self->{mesh}->[$n+$node]->{activation} = $value;
}
# Set the activation threshold for a specific layer.
# Only applicable if that layer uses "sigmoid" or "sigmoid_2"
# usage: $net->threshold($layer,$threshold);
sub threshold {
my $self = shift;
my $layer = shift || 0;
my $value = shift || 0.5;
my $n = 0;
no strict 'refs';
for(0..$layer-1){$n+=$self->{layers}->[$_]}
for($n..$n+$self->{layers}->[$layer]-1) {
$self->{mesh}->[$_]->{threshold} = $value;
}
}
# Applies a threshold to a specific node
sub node_threshold {
my $self = shift;
my $layer = shift || 0;
my $node = shift || 0;
my $value = shift || 0.5;
my $n = 0;
no strict 'refs';
for(0..$layer-1){$n+=$self->{layers}->[$_]}
$self->{mesh}->[$n+$node]->{threshold} = $value;
}
# Set mean (avg.) flag for a layer.
# usage: $net->mean($layer,$flag);
# If $flag is true, it enables finding the mean for that layer,
# If $flag is false, disables mean.
sub mean {
my $self = shift;
my $layer = shift || 0;
my $value = shift || 0;
my $n = 0;
no strict 'refs';
for(0..$layer-1){$n+=$self->{layers}->[$_]}
for($n..$n+$self->{layers}->[$layer]-1) {
$self->{mesh}->[$_]->{mean} = $value;
}
}
# Returns a pcx object
sub load_pcx {
my $self = shift;
my $file = shift;
eval('use PCX::Loader');
if(@_) {
$self->{error}="Cannot load PCX::Loader module: @_";
return undef;
}
return PCX::Loader->new($self,$file);
}
# Crunch a string of words into a map
sub crunch {
my $self = shift;
my @ws = split(/[\s\t]/,shift);
my (@map,$ic);
for my $a (0..$#ws) {
$ic=$self->crunched($ws[$a]);
if(!defined $ic) {
$self->{_crunched}->{list}->[$self->{_crunched}->{_length}++]=$ws[$a];
$map[$a]=$self->{_crunched}->{_length};
} else {
$map[$a]=$ic;
}
}
return \@map;
}
# Finds if a word has been crunched.
# Returns undef on failure, word index for success.
sub crunched {
my $self = shift;
for my $a (0..$self->{_crunched}->{_length}-1) {
return $a+1 if($self->{_crunched}->{list}->[$a] eq $_[0]);
}
$self->{error} = "Word \"$_[0]\" not found.";
return undef;
}
# Alias for crunched(), above
sub word { crunched(@_) }
# Uncrunches a map (array ref) into an array of words (not an array ref)
# and returns array
sub uncrunch {
my $self = shift;
my $map = shift;
my ($c,$el,$x);
foreach $el (@{$map}) {
$c .= $self->{_crunched}->{list}->[$el-1].' ';
}
return $c;
}
# Sets/gets randomness facter in the network. Setting a value of 0
# disables random factors.
sub random {
my $self = shift;
my $rand = shift;
return $self->{random} if(!(defined $rand));
$self->{random} = $rand;
}
# Sets/gets column width for printing lists in debug modes 1,3, and 4.
sub col_width {
my $self = shift;
my $width = shift;
return $self->{col_width} if(!$width);
$self->{col_width} = $width;
}
# Sets/gets run const. facter in the network. Setting a value of 0
# disables run const. factor.
sub const {
my $self = shift;
my $const = shift;
return $self->{const} if(!(defined $const));
$self->{const} = $const;
}
# Return benchmark time from last learn() operation.
sub benchmark {
shift->{benchmarked};
}
# Same as benchmark()
sub benchmarked {
benchmark(shift);
}
# Return the last error in the mesh, or undef if no error.
sub error {
my $self = shift;
return undef if !$self->{error};
chomp($self->{error});
return $self->{error}."\n";
}
# Used to format array ref into columns
# Usage:
# join_cols(\@array,$row_length_in_elements,$high_state_character,$low_state_character);
# Can also be called as method of your neural net.
# If $high_state_character is null, prints actual numerical values of each element.
sub join_cols {
no strict 'refs';
shift if(substr($_[0],0,4) eq 'AI::');
my $map = shift;
my $break = shift;
my $a = shift;
my $b = shift;
my $x;
foreach my $el (@{$map}) {
my $str = ((int($el))?$a:$b);
$str=$el."\0" if(!$a);
print $str; $x++;
if($x>$break-1) { print "\n"; $x=0; }
}
print "\n";
}
# Returns percentage difference between all elements of two
# array refs of exact same length (in elements).
# Now calculates actual difference in numerical value.
sub pdiff {
no strict 'refs';
shift if(substr($_[0],0,4) eq 'AI::');
my $a1 = shift;
my $a2 = shift;
my $a1s = $#{$a1};
my $a2s = $#{$a2};
my ($a,$b,$diff,$t);
$diff=0;
for my $x (0..$a1s) {
$a = $a1->[$x]; $b = $a2->[$x];
if($a!=$b) {
if($a<$b){$t=$a;$a=$b;$b=$t;}
$a=1 if(!$a); $diff+=(($a-$b)/$a)*100;
}
}
$a1s = 1 if(!$a1s);
return sprintf("%.10f",($diff/$a1s));
}
# Returns $fa as a percentage of $fb
sub p {
shift if(substr($_[0],0,4) eq 'AI::');
my ($fa,$fb)=(shift,shift);
sprintf("%.3f",$fa/$fb*100); #((($fb-$fa)*((($fb-$fa)<0)?-1:1))/$fa)*100
}
# Returns the index of the element in array REF passed with the highest
# comparative value
sub high {
shift if(substr($_[0],0,4) eq 'AI::');
my $ref1 = shift; my ($el,$len,$tmp); $tmp=0;
foreach $el (@{$ref1}) { $len++ }
for my $x (0..$len-1) { $tmp = $x if($ref1->[$x] > $ref1->[$tmp]) }
return $tmp;
}
# Returns the index of the element in array REF passed with the lowest
# comparative value
sub low {
shift if(substr($_[0],0,4) eq 'AI::');
my $ref1 = shift; my ($el,$len,$tmp); $tmp=0;
foreach $el (@{$ref1}) { $len++ }
for my $x (0..$len-1) { $tmp = $x if($ref1->[$x] < $ref1->[$tmp]) }
return $tmp;
}
# Following is a collection of a few nifty custom activation functions.
# range() is exported by default, the rest you can get with:
# use AI::NeuralNet::Mesh ':acts'
# The ':all' tag also gets these into your namespace.
#
# range() returns a closure limiting the output
# of that node to a specified set of values.
# Good for output layers.
#
# usage example:
# $net->activation(4,range(0..5));
# or:
# ..
# {
# nodes => 1,
# activation => range 5..2
# }
# ..
# You can also pass an array containing the range
# values (not array ref), or you can pass a comma-
# seperated list of values as parameters:
#
# $net->activation(4,range(@numbers));
# $net->activation(4,range(6,15,26,106,28,3));
#
# Note: when using a range() activatior, train the
# net TWICE on the data set, because the first time
# the range() function searches for the top value in
# the inputs, and therefore, results could flucuate.
# The second learning cycle guarantees more accuracy.
#
sub range {
my @r=@_;
sub{$_[1]->{t}=$_[0]if($_[0]>$_[1]->{t});$r[intr($_[0]/$_[1]->{t}*$#r)]}
}
#
# ramp() preforms smooth ramp activation between 0 and 1 if $r is 1,
# or between -1 and 1 if $r is 2. $r defaults to 1, as you can see.
#
# Note: when using a ramp() activatior, train the
# net at least TWICE on the data set, because the first
# time the ramp() function searches for the top value in
# the inputs, and therefore, results could flucuate.
# The second learning cycle guarantees more accuracy.
#
sub ramp {
my $r=shift||1;my $t=($r<2)?0:-1;
sub{$_[1]->{t}=$_[0]if($_[0]>$_[1]->{t});$_[0]/$_[1]->{t}*$r-$b}
}
# Self explanitory, pretty much. $threshold is used to decide if an input
# is true or false (1 or 0). If an input is below $threshold, it is false.
sub and_gate {
my $threshold = shift || 0.5;
sub {
my $sum = shift;
my $self = shift;
for my $x (0..$self->{_inputs_size}-1) { return $self->{_parent}->{const} if!$self->{_inputs}->[$x]->{value}<$threshold }
return $sum/$self->{_inputs_size};
}
}
# Self explanitory, $threshold is used same as above.
sub or_gate {
my $threshold = shift || 0.5;
sub {
my $sum = shift;
my $self = shift;
for my $x (0..$self->{_inputs_size}-1) { return $sum/$self->{_inputs_size} if!$self->{_inputs}->[$x]->{value}<$threshold }
return $self->{_parent}->{const};
}
}
1;
package AI::NeuralNet::Mesh::node;
use strict;
# Node constructor
sub new {
my $type = shift;
my $self ={
_parent => shift,
_inputs => [],
_outputs => []
};
bless $self, $type;
}
# Receive inputs from other nodes, and also send
# outputs on.
sub input {
my $self = shift;
my $input = shift;
my $from_id = shift;
$self->{_inputs}->[$from_id]->{value} = $input * $self->{_inputs}->[$from_id]->{weight};
$self->{_inputs}->[$from_id]->{input} = $input;
$self->{_inputs}->[$from_id]->{fired} = 1;
$self->{_parent}->d("got input $input from id $from_id, weighted to $self->{_inputs}->[$from_id]->{value}.\n",1);
my $flag = 1;
for my $x (0..$self->{_inputs_size}-1) { $flag = 0 if(!$self->{_inputs}->[$x]->{fired}) }
if ($flag) {
$self->{_parent}->d("all inputs fired for $self.\n",1);
my $output = 0;
# Sum
for my $i (@{$self->{_inputs}}) {
$output += $i->{value};
}
# Handle activations, thresholds, and means
$output /= $self->{_inputs_size} if($self->{flag_mean});
#$output += (rand()*$self->{_parent}->{random});
$output = ($output>=$self->{threshold})?1:0 if(($self->{activation} eq "sigmoid") || ($self->{activation} eq "sigmoid_1"));
if($self->{activation} eq "sigmoid_2") {
$output = 1 if($output >$self->{threshold});
$output = -1 if($output <$self->{threshold});
$output = 0 if($output==$self->{threshold});
}
# Handle CODE refs
$output = &{$self->{activation}}($output,$self) if(ref($self->{activation}) eq "CODE");
# Send output
for my $o (@{$self->{_outputs}}) { $o->{node}->input($output,$o->{from_id}) }
} else {
$self->{_parent}->d("all inputs have NOT fired for $self.\n",1);
}
}
sub add_input_node {
my $self = shift;
my $node = shift;
my $i = $self->{_inputs_size} || 0;
$self->{_inputs}->[$i]->{node} = $node;
$self->{_inputs}->[$i]->{value} = 0;
$self->{_inputs}->[$i]->{weight} = 1; #rand()*1;
$self->{_inputs}->[$i]->{fired} = 0;
$self->{_inputs_size} = ++$i;
return $i-1;
}
sub add_output_node {
my $self = shift;
my $node = shift;
my $i = $self->{_outputs_size} || 0;
$self->{_outputs}->[$i]->{node} = $node;
$self->{_outputs}->[$i]->{from_id} = $node->add_input_node($self);
$self->{_outputs_size} = ++$i;
return $i-1;
}
sub adjust_weight {
my $self = shift;
my $inc = shift;
for my $i (@{$self->{_inputs}}) {
$i->{weight} += $inc * $i->{weight};
$i->{node}->adjust_weight($inc) if($i->{node});
}
}
1;
# Internal usage, prevents recursion on empty nodes.
package AI::NeuralNet::Mesh::cap;
sub new { bless {}, shift }
sub input {}
sub adjust_weight {}
sub add_output_node {}
sub add_input_node {}
1;
# Internal usage, collects data from output layer.
package AI::NeuralNet::Mesh::output;
use strict;
sub new {
my $type = shift;
my $self ={
_parent => shift,
_inputs => [],
};
bless $self, $type;
}
sub add_input_node {
my $self = shift;
return (++$self->{_inputs_size})-1;
}
sub input {
my $self = shift;
my $input = shift;
my $from_id = shift;
$self->{_parent}->d("GOT INPUT [$input] FROM [$from_id]\n",1);
$self->{_inputs}->[$from_id] = $self->{_parent}->intr($input);
}
sub get_outputs {
my $self = shift;
return $self->{_inputs};
}
1;
__END__
=head1 NAME
AI::NeuralNet::Mesh - An optimized, accurate neural network Mesh.
=head1 SYNOPSIS
use AI::NeuralNet::Mesh;
# Create a mesh with 2 layers, 2 nodes/layer, and one output node.
my $net = new AI::NeuralNet::Mesh(2,2,1);
# Teach the network the AND function
$net->learn([0,0],[0]);
$net->learn([0,1],[0]);
$net->learn([1,0],[0]);
$net->learn([1,1],[1]);
# Present it with two test cases
my $result_bit_1 = $net->run([0,1])->[0];
my $result_bit_2 = $net->run([1,1])->[0];
# Display the results
print "AND test with inputs (0,1): $result_bit_1\n";
print "AND test with inputs (1,1): $result_bit_2\n";
=head1 VERSION & UPDATES
This is version B<0.44>, an update release for version 0.43.
This fixed the usage conflict with perl 5.3.3.
With this version I have gone through and tuned up many area
of this module, including the descent algorithim in learn(),
as well as four custom activation functions, and several export
tag sets. With this release, I have also included a few
new and more practical example scripts. (See ex_wine.pl) This release
for details. This version also fixes a big bug that I never knew about
until writing some demos for this version - that is, when trying to use
more than one output node, the mesh would freeze in learning. But, that
is fixed now, and you can have as many outputs as you want (how does 3
inputs and 50 outputs sound? :-)
=head1 DESCRIPTION
AI::NeuralNet::Mesh is an optimized, accurate neural network Mesh.
It was designed with accruacy and speed in mind.
This network model is very flexable. It will allow for clasic binary
operation or any range of integer or floating-point inputs you care
to provide. With this you can change activation types on a per node or
per layer basis (you can even include your own anonymous subs as
activation types). You can add sigmoid transfer functions and control
the threshold. You can learn data sets in batch, and load CSV data
set files. You can do almost anything you need to with this module.
This code is deigned to be flexable. Any new ideas for this module?
See AUTHOR, below, for contact info.
This module is designed to also be a customizable, extensable
neural network simulation toolkit. Through a combination of setting
the $Connection variable and using custom activation functions, as
well as basic package inheritance, you can simulate many different
types of neural network structures with very little new code written
by you.
In this module I have included a more accurate form of "learning" for the
mesh. This form preforms descent toward a local error minimum (0) on a
directional delta, rather than the desired value for that node. This allows
for better, and more accurate results with larger datasets. This module also
uses a simpler recursion technique which, suprisingly, is more accurate than
the original technique that I've used in other ANNs.
=head1 EXPORTS
This module exports three functions by default:
range
intr
pdiff
See range() intr() and pdiff() for description of their respective functions.
Also provided are several export tag sets for usage in the form of:
use AI::NeuralNet::Mesh ':tag';
Tag sets are:
:default
- These functions are always exported.
- Exports:
range()
intr()
pdiff()
:all
- Exports:
p()
high()
low()
range()
ramp()
and_gate()
or_gate()
:p
- Exports:
p()
high()
low()
:acts
- Exports:
ramp()
and_gate()
or_gate()
See the respective methods/functions for information about
each method/functions usage.
=head1 METHODS
=item AI::NeuralNet::Mesh->new();
There are four ways to construct a new network with new(). Each is detailed below.
P.S. Don't worry, the old C<new($layers, $nodes [, $outputs])> still works like always!
=item AI::NeuralNet::Mesh->new($layers, $nodes [, $outputs]);
Returns a newly created neural network from an C<AI::NeuralNet::Mesh>
object. The network will have C<$layers> number of layers in it
and it will have C<$nodes> number of nodes per layer.
There is an optional parameter of $outputs, which specifies the number
of output neurons to provide. If $outputs is not specified, $outputs
defaults to equal $size.
=item AI::NeuralNet::Mesh->new($file);
This will automatically create a new network from the file C<$file>. It will
return undef if the file was of an incorrect format or non-existant. Otherwise,
it will return a blessed refrence to a network completly restored from C<$file>.
=item AI::NeuralNet::Mesh->new(\@layer_sizes);
This constructor will make a network with the number of layers corresponding to the length
in elements of the array ref passed. Each element in the array ref passed is expected
to contain an integer specifying the number of nodes (neurons) in that layer. The first
layer ($layer_sizes[0]) is to be the input layer, and the last layer in @layer_sizes is to be
the output layer.
Example:
my $net = AI::NeuralNet::Mesh->new([2,3,1]);
Creates a network with 2 input nodes, 3 hidden nodes, and 1 output node.
=item AI::NeuralNet::Mesh->new(\@array_of_hashes);
Another dandy constructor...this is my favorite. It allows you to tailor the number of layers,
the size of the layers, the activation type (you can even add anonymous inline subs with this one),
and even the threshold, all with one array ref-ed constructor.
Example:
my $net = AI::NeuralNet::Mesh->new([
{
nodes => 2,
activation => linear
},
{
nodes => 3,
activation => sub {
my $sum = shift;
return $sum + rand()*1;
}
},
{
nodes => 1,
activation => sigmoid,
threshold => 0.75
}
]);
Interesting, eh? What you are basically passing is this:
my @info = (
{ },
{ },
{ },
...
);
You are passing an array ref who's each element is a hash refrence. Each
hash refrence, or more precisely, each element in the array refrence you are passing
to the constructor, represents a layer in the network. Like the constructor above,
the first element is the input layer, and the last is the output layer. The rest are
hidden layers.
Each hash refrence is expected to have AT LEAST the "nodes" key set to the number
of nodes (neurons) in that layer. The other two keys are optional. If "activation" is left
out, it defaults to "linear". If "threshold" is left out, it defaults to 0.50.
The "activation" key can be one of four values:
linear ( simply use sum of inputs as output )
sigmoid [ sigmoid_1 ] ( only positive sigmoid )
sigmoid_2 ( positive / 0 /negative sigmoid )
\&code_ref;
"sigmoid_1" is an alias for "sigmoid".
The code ref option allows you to have a custom activation function for that layer.
The code ref is called with this syntax:
$output = &$code_ref($sum_of_inputs, $self);
The code ref is expected to return a value to be used as the output of the node.
The code ref also has access to all the data of that node through the second argument,
a blessed hash refrence to that node.
See CUSTOM ACTIVATION FUNCTIONS for information on several included activation functions
other than the ones listed above.
Three of the activation syntaxes are shown in the first constructor above, the "linear",
"sigmoid" and code ref types.
You can also set the activation and threshold values after network creation with the
activation() and threshold() methods.
=item $net->learn($input_map_ref, $desired_result_ref [, options ]);
NOTE: learn_set() now has increment-degrading turned OFF by default. See note
on the degrade flag, below.
This will 'teach' a network to associate an new input map with a desired
result. It will return a string containg benchmarking information.
You can also specify strings as inputs and ouputs to learn, and they will be
crunched automatically. Example:
$net->learn('corn', 'cob');
Note, the old method of calling crunch on the values still works just as well.
The first two arguments may be array refs (or now, strings), and they may be
of different lengths.
Options should be written on hash form. There are three options:
inc => $learning_gradient
max => $maximum_iterations
error => $maximum_allowable_percentage_of_error
degrade => $degrade_increment_flag
$learning_gradient is an optional value used to adjust the weights of the internal
connections. If $learning_gradient is ommitted, it defaults to 0.002.
$maximum_iterations is the maximum numbers of iteration the loop should do.
It defaults to 1024. Set it to 0 if you never want the loop to quit before
the pattern is perfectly learned.
$maximum_allowable_percentage_of_error is the maximum allowable error to have. If
this is set, then learn() will return when the perecentage difference between the
actual results and desired results falls below $maximum_allowable_percentage_of_error.
If you do not include 'error', or $maximum_allowable_percentage_of_error is set to -1,
then learn() will not return until it gets an exact match for the desired result OR it
reaches $maximum_iterations.
$degrade_increment_flag is a simple flag used to allow/dissalow increment degrading
during learning based on a product of the error difference with several other factors.
$degrade_increment_flag is off by default. Setting $degrade_increment_flag to a true
value turns increment degrading on.
In previous module releases $degrade_increment_flag was not used, as increment degrading
was always on. In this release I have looked at several other network types as well
as several texts and decided that it would be better to not use increment degrading. The
option is still there for those that feel the inclination to use it. I have found some areas
that do need the degrade flag to work at a faster speed. See test.pl for an example. If
the degrade flag wasn't in test.pl, it would take a very long time to learn.
=item $net->learn_set(\@set, [ options ]);
This takes the same options as learn() (learn_set() uses learn() internally)
and allows you to specify a set to learn, rather than individual patterns.
A dataset is an array refrence with at least two elements in the array,
each element being another array refrence (or now, a scalar string). For
each pattern to learn, you must specify an input array ref, and an ouput
array ref as the next element. Example:
my @set = (
# inputs outputs
[ 1,2,3,4 ], [ 1,3,5,6 ],
[ 0,2,5,6 ], [ 0,2,1,2 ]
);
Inputs and outputs in the dataset can also be strings.
See the paragraph on measuring forgetfulness, below. There are
two learn_set()-specific option tags available:
flag => $flag
pattern => $row
If "flag" is set to some TRUE value, as in "flag => 1" in the hash of options, or if the option "flag"
is not set, then it will return a percentage represting the amount of forgetfullness. Otherwise,
learn_set() will return an integer specifying the amount of forgetfulness when all the patterns
are learned.
If "pattern" is set, then learn_set() will use that pattern in the data set to measure forgetfulness by.
If "pattern" is omitted, it defaults to the first pattern in the set. Example:
my @set = (
[ 0,1,0,1 ], [ 0 ],
[ 0,0,1,0 ], [ 1 ],
[ 1,1,0,1 ], [ 2 ], # <---
[ 0,1,1,0 ], [ 3 ]
);
If you wish to measure forgetfulness as indicated by the line with the arrow, then you would
pass 2 as the "pattern" option, as in "pattern => 2".
Now why the heck would anyone want to measure forgetfulness, you ask? Maybe you wonder how I
even measure that. Well, it is not a vital value that you have to know. I just put in a
"forgetfulness measure" one day because I thought it would be neat to know.
How the module measures forgetfulness is this: First, it learns all the patterns
in the set provided, then it will run the very first pattern (or whatever pattern
is specified by the "row" option) in the set after it has finished learning. It
will compare the run() output with the desired output as specified in the dataset.
In a perfect world, the two should match exactly. What we measure is how much that
they don't match, thus the amount of forgetfulness the network has.
Example (from examples/ex_dow.pl):
# Data from 1989 (as far as I know..this is taken from example data on BrainMaker)
my @data = (
# Mo CPI CPI-1 CPI-3 Oil Oil-1 Oil-3 Dow Dow-1 Dow-3 Dow Ave (output)
[ 1, 229, 220, 146, 20.0, 21.9, 19.5, 2645, 2652, 2597], [ 2647 ],
[ 2, 235, 226, 155, 19.8, 20.0, 18.3, 2633, 2645, 2585], [ 2637 ],
[ 3, 244, 235, 164, 19.6, 19.8, 18.1, 2627, 2633, 2579], [ 2630 ],
[ 4, 261, 244, 181, 19.6, 19.6, 18.1, 2611, 2627, 2563], [ 2620 ],
[ 5, 276, 261, 196, 19.5, 19.6, 18.0, 2630, 2611, 2582], [ 2638 ],
[ 6, 287, 276, 207, 19.5, 19.5, 18.0, 2637, 2630, 2589], [ 2635 ],
[ 7, 296, 287, 212, 19.3, 19.5, 17.8, 2640, 2637, 2592], [ 2641 ]
);
# Learn the set
my $f = $net->learn_set(\@data,
inc => 0.1,
max => 500,
);
# Print it
print "Forgetfullness: $f%";
This is a snippet from the example script examples/finance.pl, which demonstrates DOW average
prediction for the next month. A more simple set defenition would be as such:
my @data = (
[ 0,1 ], [ 1 ],
[ 1,0 ], [ 0 ]
);
$net->learn_set(\@data);
Same effect as above, but not the same data (obviously).
=item $net->run($input_map_ref);
This method will apply the given array ref at the input layer of the neural network, and
it will return an array ref to the output of the network. run() will now automatically crunch()
a string given as an input (See the crunch() method for info on crunching).
Example Usage:
my $inputs = [ 1,1,0,1 ];
my $outputs = $net->run($inputs);
You can also do this with a string:
my $outputs = $net->run('cloudy - wind is 5 MPH NW');
See also run_uc() and run_set() below.
=item $net->run_uc($input_map_ref);
This method does the same thing as this code:
$net->uncrunch($net->run($input_map_ref));
All that run_uc() does is that it automatically calls uncrunch() on the output, regardless
of whether the input was crunch() -ed or not.
=item $net->run_set($set);
This takes an array ref of the same structure as the learn_set() method, above. It returns
an array ref. Each element in the returned array ref represents the output for the corresponding
element in the dataset passed. Uses run() internally.
=item $net->get_outs($set);
Simple utility function which takes an array ref of the same structure as the learn_set() method,
above. It returns an array ref of the same type as run_set() wherein each element contains an
output value. The output values are the target values specified in the $set passed. Each element
in the returned array ref represents the output value for the corrseponding row in the dataset
passed. (A row is two elements of the dataset together, see learn_set() for dataset structure.)
=item $net->load_set($file,$column,$seperator);
Loads a CSV-like dataset from disk
Returns a data set of the same structure as required by the
learn_set() method. $file is the disk file to load set from.
$column an optional variable specifying the column in the
data set to use as the class attribute. $class defaults to 0.
$seperator is an optional variable specifying the seperator
character between values. $seperator defaults to ',' (a single comma).
NOTE: This does not handle quoted fields, or any other record
seperator other than "\n".
The returned array ref is suitable for passing directly to
learn_set() or get_outs().
=item $net->range();
See CUSTOM ACTIVATION FUNCTIONS for information on several included activation functions.
=item $net->benchmark();
=item $net->benchmarked();
This returns a benchmark info string for the last learn() call.
It is easily printed as a string, as following:
print "Last learn() took ",$net->benchmark(),"\n";
=item $net->verbose($level);
=item $net->verbosity($level);
=item $net->v($level);
=item $net->debug($level)
Note: verbose(), verbosity(), and v() are all functional aliases for debug().
Toggles debugging off if called with $level = 0 or no arguments. There are several levels
of debugging.
NOTE: Debugging verbosity has been toned down somewhat from AI::NeuralNet::BackProp,
but level 4 still prints the same amount of information as you were used to. The other
levels, however, are mostly for advanced use. Not much explanation in the other
levels, but they are included for those of you that feel daring (or just plain bored.)
Level 0 ($level = 0) : Default, no debugging information printed. All printing is
left to calling script.
Level 1 ($level = 1) : Displays the activity between nodes, prints what values were
received and what they were weighted to.
Level 2 ($level = 2) : Just prints info from the learn() loop, in the form of "got: X, wanted Y"
type of information. This is about the third most useful debugging level, after level 12 and
level 4.
Level 3 ($level = 3) : I don't think I included any level 3 debugs in this version.
Level 4 ($level = 4) : This level is the one I use most. It is only used during learning. It
displays the current error (difference between actual outputs and the target outputs you
asked for), as well as the current loop number and the benchmark time for the last learn cycle.
Also printed are the actual outputs and the target outputs below the benchmark times.
Level 12 ($level = 12) : Level 12 prints a dot (period) [.] after each learning loop is
complete. This is useful for letting the user know that stuff is happening, but without
having to display any of the internal variables. I use this in the ex_aln.pl demo,
as well as the ex_agents.pl demo.
Toggles debuging off when called with no arguments.
=item $net->save($filename);
This will save the complete state of the network to disk, including all weights and any
words crunched with crunch() . Also saves the layer size and activations of the network.
NOTE: The only activation type NOT saved is the CODE ref type, which must be set again
after loading.
This uses a simple flat-file text storage format, and therefore the network files should
be fairly portable.
This method will return undef if there was a problem with writing the file. If there is an
error, it will set the internal error message, which you can retrive with the error() method,
below.
If there were no errors, it will return a refrence to $net.
=item $net->load($filename);
This will load from disk any network saved by save() and completly restore the internal
state at the point it was save() was called at.
If the file is of an invalid file type, then load() will
return undef. Use the error() method, below, to print the error message.
If there were no errors, it will return a refrence to $net.
UPDATE: $filename can now be a newline-seperated set of mesh data. This enables you
to do $net->load(join("\n",<DATA>)) and other fun things. I added this mainly
for a demo I'm writing but not qutie done with yet. So, Cheers!
=item $net->activation($layer,$type);
This sets the activation type for layer C<$layer>.
C<$type> can be one of four values:
linear ( simply use sum of inputs as output )
sigmoid [ sigmoid_1 ] ( only positive sigmoid )
sigmoid_2 ( positive / 0 /negative sigmoid )
\&code_ref;
"sigmoid_1" is an alias for "sigmoid".
The code ref option allows you to have a custom activation function for that layer.
The code ref is called with this syntax:
$output = &$code_ref($sum_of_inputs, $self);
The code ref is expected to return a value to be used as the output of the node.
The code ref also has access to all the data of that node through the second argument,
a blessed hash refrence to that node.
See CUSTOM ACTIVATION FUNCTIONS for information on several included activation functions
other than the ones listed above.
The activation type for each layer is preserved across load/save calls.
EXCEPTION: Due to the constraints of Perl, I cannot load/save the actual subs that the code
ref option points to. Therefore, you must re-apply any code ref activation types after a
load() call.
=item $net->node_activation($layer,$node,$type);
This sets the activation function for a specific node in a layer. The same notes apply
here as to the activation() method above.
=item $net->threshold($layer,$value);
This sets the activation threshold for a specific layer. The threshold only is used
when activation is set to "sigmoid", "sigmoid_1", or "sigmoid_2".
=item $net->node_threshold($layer,$node,$value);
This sets the activation threshold for a specific node in a layer. The threshold only is used
when activation is set to "sigmoid", "sigmoid_1", or "sigmoid_2".
=item $net->join_cols($array_ref,$row_length_in_elements,$high_state_character,$low_state_character);
This is more of a utility function than any real necessary function of the package.
Instead of joining all the elements of the array together in one long string, like join() ,
it prints the elements of $array_ref to STDIO, adding a newline (\n) after every $row_length_in_elements
number of elements has passed. Additionally, if you include a $high_state_character and a $low_state_character,
it will print the $high_state_character (can be more than one character) for every element that
has a true value, and the $low_state_character for every element that has a false value.
If you do not supply a $high_state_character, or the $high_state_character is a null or empty or
undefined string, it join_cols() will just print the numerical value of each element seperated
by a null character (\0). join_cols() defaults to the latter behaviour.
=item $net->extend(\@array_of_hashes);
This allows you to re-apply any activations and thresholds with the same array ref which
you created a network with. This is useful for re-applying code ref activations after a load()
call without having to type the code ref twice.
You can also specify the extension in a simple array ref like this:
$net->extend([2,3,1]);
Which will simply add more nodes if needed to set the number of nodes in each layer to their
respective elements. This works just like the respective new() constructor, above.
NOTE: Your net will probably require re-training after adding nodes.
=item $net->extend_layer($layer,\%hash);
With this you can modify only one layer with its specifications in a hash refrence. This hash
refrence uses the same keys as for the last new() constructor form, above.
You can also specify just the number of nodes for the layer in this form:
$net->extend_layer(0,5);
Which will set the number of nodes in layer 0 to 5 nodes. This is the same as calling:
$net->add_nodes(0,5);
Which does the exact same thing. See add_nodes() below.
NOTE: Your net will probably require re-training after adding nodes.
=item $net->add_nodes($layer,$total_nodes);
This method was created mainly to service the extend*() group of functions, but it
can also be called independently. This will add nodes as needed to layer C<$layer> to
make the nodes in layer equal to $total_nodes.
NOTE: Your net will probably require re-training after adding nodes.
=item $net->p($a,$b);
Returns a floating point number which represents $a as a percentage of $b.
=item $net->intr($float);
Rounds a floating-point number rounded to an integer using sprintf() and int() , Provides
better rounding than just calling int() on the float. Also used very heavily internally.
=item $net->high($array_ref);
Returns the index of the element in array REF passed with the highest comparative value.
=item $net->low($array_ref);
Returns the index of the element in array REF passed with the lowest comparative value.
=item $net->pdiff($array_ref_A, $array_ref_B);
This function is used VERY heavily internally to calculate the difference in percent
between elements of the two array refs passed. It returns a %.20f (sprintf-format)
percent sting.
=item $net->show();
This will dump a simple listing of all the weights of all the connections of every neuron
in the network to STDIO.
=item $net->crunch($string);
This splits a string passed with /[\s\t]/ into an array ref containing unique indexes
to the words. The words are stored in an intenal array and preserved across load() and save()
calls. This is designed to be used to generate unique maps sutible for passing to learn() and
run() directly. It returns an array ref.
The words are not duplicated internally. For example:
$net->crunch("How are you?");
Will probably return an array ref containing 1,2,3. A subsequent call of:
$net->crunch("How is Jane?");
Will probably return an array ref containing 1,4,5. Notice, the first element stayed
the same. That is because it already stored the word "How". So, each word is stored
only once internally and the returned array ref reflects that.
=item $net->uncrunch($array_ref);
Uncrunches a map (array ref) into an scalar string of words seperated by ' ' and returns the
string. This is ment to be used as a counterpart to the crunch() method, above, possibly to
uncrunch() the output of a run() call. Consider the below code (also in ./examples/ex1.pl):
use AI::NeuralNet::Mesh;
my $net = AI::NeuralNet::Mesh->new(2,3);
for (0..3) {
$net->learn_set([
$net->crunch("I love chips."), $net->crunch("That's Junk Food!")),
$net->crunch("I love apples."), $net->crunch("Good, Healthy Food.")),
$net->crunch("I love pop."), $net->crunch("That's Junk Food!")),
$net->crunch("I love oranges."),$net->crunch("Good, Healthy Food."))
]);
}
print $net->run_uc("I love corn.")),"\n";
On my system, this responds with, "Good, Healthy Food." If you try to run crunch() with
"I love pop.", though, you will probably get "Food! apples. apples." (At least it returns
that on my system.) As you can see, the associations are not yet perfect, but it can make
for some interesting demos!
=item $net->crunched($word);
This will return undef if the word is not in the internal crunch list, or it will return the
index of the word if it exists in the crunch list.
If the word is not in the list, it will set the internal error value with a text message
that you can retrive with the error() method, below.
=item $net->word($word);
A function alias for crunched().
=item $net->col_width($width);
This is useful for formating the debugging output of Level 4 if you are learning simple
bitmaps. This will set the debugger to automatically insert a line break after that many
elements in the map output when dumping the currently run map during a learn loop.
It will return the current width when called with a 0 or undef value.
The column width is preserved across load() and save() calls.
=item $net->random($rand);
This will set the randomness factor from the network. Default is 0. When called
with no arguments, or an undef value, it will return current randomness value. When
called with a 0 value, it will disable randomness in the network. The randomness factor
is preserved across load() and save() calls.
=item $net->const($const);
This sets the run const. for the network. The run const. is a value that is added
to every input line when a set of inputs are run() or learn() -ed, to prevent the
network from hanging on a 0 value. When called with no arguments, it returns the current
const. value. It defaults to 0.0001 on a newly-created network. The run const. value
is preserved across load() and save() calls.
=item $net->error();
Returns the last error message which occured in the mesh, or undef if no errors have
occured.
=item $net->load_pcx($filename);
NOTE: To use this function, you must have PCX::Loader installed. If you do not have
PCX::Loader installed, it will return undef and store an error for you to retrive with
the error() method, below.
This is a treat... this routine will load a PCX-format file (yah, I know ... ancient
format ... but it is the only one I could find specs for to write it in Perl. If
anyone can get specs for any other formats, or could write a loader for them, I
would be very grateful!) Anyways, a PCX-format file that is exactly 320x200 with 8 bits
per pixel, with pure Perl. It returns a blessed refrence to a PCX::Loader object, which
supports the following routinges/members. See example files ex_pcx.pl and ex_pcxl.pl in
the ./examples/ directory.
See C<perldoc PCX::Loader> for information on the methods of the object returned.
You can download PCX::Loader from
http://www.josiah.countystart.com/modules/get.pl?pcx-loader:mpod
=head1 CUSTOM ACTIVATION FUNCTIONS
Included in this package are four custom activation functions meant to be used
as a guide to create your own, as well as to be useful to you in normal use of the
module. There is only one function exported by default into your namespace, which
is the range() functions. These are not meant to be used as methods, but as functions.
These functions return code refs to a Perl closure which does the actual work when
the time comes.
=item range(0..X);
=item range(@range);
=item range(A,B,C);
range() returns a closure limiting the output
of that node to a specified set of values.
Good for use in output layers.
Usage example:
$net->activation(4,range(0..5));
or (in the new() hash constructor form):
..
{
nodes => 1,
activation => range 5..2
}
..
You can also pass an array containing the range
values (not array ref), or you can pass a comma-
seperated list of values as parameters:
$net->activation(4,range(@numbers));
$net->activation(4,range(6,15,26,106,28,3));
Note: when using a range() activatior, train the
net TWICE on the data set, because the first time
the range() function searches for the top value in
the inputs, and therefore, results could flucuate.
The second learning cycle guarantees more accuracy.
The actual code that implements the range closure is
a bit convulted, so I will expand on it here as a simple
tutorial for custom activation functions.
= line 1 = sub {
= line 2 = my @values = ( 6..10 );
= line 3 = my $sum = shift;
= line 4 = my $self = shift;
= line 5 = $self->{top_value}=$sum if($sum>$self->{top_value});
= line 6 = my $index = intr($sum/$self->{top_value}*$#values);
= line 7 = return $values[$index];
= line 8 = }
Now, the actual function fits in one line of code, but I expanded it a bit
here. Line 1 creates our array of allowed output values. Lines two and
three grab our parameters off the stack which allow us access to the
internals of this node. Line 5 checks to see if the sum output of this
node is higher than any previously encountered, and, if so, it sets
in the ramp() activator. Line 6 computes the index into the allowed
values array by first scaling the $sum to be between 0 and 1 and then
expanding it to fit smoothly inside the number of elements in the array. Then
we simply round to an integer and pluck that index from the array and
use it as the output value for that node.
See? It's not that hard! Using custom activation functions, you could do
just about anything with the node that you want to, since you have
access to the node just as if you were a blessed member of that node's object.
=item ramp($r);
ramp() preforms smooth ramp activation between 0 and 1 if $r is 1,
or between -1 and 1 if $r is 2. $r defaults to 1.
You can get this into your namespace with the ':acts' export
tag as so:
use AI::NeuralNet::Mesh ':acts';
Note: when using a ramp() activatior, train the
net at least TWICE on the data set, because the first
time the ramp() function searches for the top value in
the inputs, and therefore, results could flucuate.
The second learning cycle guarantees more accuracy.
No code to show here, as it is almost exactly the same as range().
=item and_gate($threshold);
Self explanitory, pretty much. This turns the node into a basic AND gate.
$threshold is used to decide if an input is true or false (1 or 0). If
an input is below $threshold, it is false. $threshold defaults to 0.5.
You can get this into your namespace with the ':acts' export
tag as so:
use AI::NeuralNet::Mesh ':acts';
Let's look at the code real quick, as it shows how to get at the indivudal
input connections:
= line 1 = sub {
= line 2 = my $sum = shift;
= line 3 = my $self = shift;
= line 4 = my $threshold = 0.50;
= line 5 = for my $x (0..$self->{_inputs_size}-1) {
= line 6 = return 0.000001 if(!$self->{_inputs}->[$x]->{value}<$threshold)
= line 7 = }
= line 8 = return $sum/$self->{_inputs_size};
= line 9 = }
Line 2 and 3 pulls in our sum and self refrence. Line 5 opens a loop to go over
all the input lines into this node. Line 6 looks at each input line's value
and comparse it to the threshold. If the value of that line is below threshold, then
we return 0.000001 to signify a 0 value. (We don't return a 0 value so that the network
doen't get hung trying to multiply a 0 by a huge weight during training [it just will
keep getting a 0 as the product, and it will never learn]). Line 8 returns the mean
value of all the inputs if all inputs were above threshold.
Very simple, eh? :)
=item or_gate($threshold)
Self explanitory. Turns the node into a basic OR gate, $threshold is used same as above.
You can get this into your namespace with the ':acts' export
tag as so:
use AI::NeuralNet::Mesh ':acts';
=head1 VARIABLES
=item $AI::NeuralNet::Mesh::Connector
This is an option is step up from average use of this module. This variable
should hold the fully qualified name of the function used to make the actual connections
between the nodes in the network. This contains '_c' by default, but if you use
this variable, be sure to add the fully qualified name of the method. For example,
in the ALN example, I use a connector in the main package called tree() instead of
the default connector. Before I call the new() constructor, I use this line of code:
$AI::NeuralNet::Mesh::Connector = 'main::tree'
The tree() function is called as a blessed method when it is used internally, providing
access to the bless refrence in the first argument. See notes on CUSTOM NETWORK CONNECTORS,
below, for more information on creating your own custom connector.
=item $AI::NeuralNet::Mesh::DEBUG
This variable controls the verbosity level. It will not hurt anything to set this
directly, yet most people find it easier to set it using the debug() method, or
any of its aliases.
=head1 CUSTOM NETWORK CONNECTORS
Creating custom network connectors is step up from average use of this module.
However, it can be very useful in creating other styles of neural networks, other
than the default fully-connected feed-foward network.
You create a custom connector by setting the variable $AI::NeuralNet::Mesh::Connector
to the fully qualified name of the function used to make the actual connections
between the nodes in the network. This variable contains '_c' by default, but if you use
this variable, be sure to add the fully qualified name of the method. For example,
in the ALN example, I use a connector in the main package called tree() instead of
the default connector. Before I call the new() constructor, I use this line of code:
$AI::NeuralNet::Mesh::Connector = 'main::tree'
The tree() function is called as a blessed method when it is used internally, providing
access to the bless refrence in the first argument.
Example connector:
sub connect_three {
my $self = shift;
my $r1a = shift;
my $r1b = shift;
my $r2a = shift;
my $r2b = shift;
my $mesh = $self->{mesh};
for my $y (0..($r1b-$r1a)-1) {
$mesh->[$y+$r1a]->add_output_node($mesh->[$y+$r2a-1]) if($y>0);
$mesh->[$y+$r1a]->add_output_node($mesh->[$y+$r2a]) if($y<($r2b-$r2a));
$mesh->[$y+$r1a]->add_output_node($mesh->[$y+$r2a+1]) if($y<($r2b-$r2a));
}
}
This is a very simple example. It feeds the outputs of every node in the first layer
to the node directly above it, as well as the nodes on either side of the node directly
above it, checking for range sides, of course.
The network is stored internally as one long array of node objects. The goal here
is to connect one range of nodes in that array to another range of nodes. The calling
function has already calculated the indices into the array, and it passed it to you
as the four arguments after the $self refrence. The first two arguments we will call
$r1a and $r1b. These define the start and end indices of the first range, or "layer." Likewise,
the next two arguemnts, $r2a and $r2b, define the start and end indices of the second
layer. We also grab a refrence to the mesh array so we dont have to type the $self
refrence over and over.
The loop that folows the arguments in the above example is very simple. It opens
a for() loop over the range of numbers, calculating the size instead of just going
$r1a..$r1b because we use the loop index with the next layer up as well.
$y + $r1a give the index into the mesh array of the current node to connect the output FROM.
We need to connect this nodes output lines to the next layers input nodes. We do this
with a simple method of the outputing node (the node at $y+$r1a), called add_output_node().
add_output_node() takes one simple arguemnt: A blessed refrence to a node that it is supposed
to output its final value TO. We get this blessed refrence with more simple addition.
$y + $r2a gives us the node directly above the first node (supposedly...I'll get to the "supposedly"
part in a minute.) By adding or subtracting from this number we get the neighbor nodes.
In the above example you can see we check the $y index to see that we havn't come close to
any of the edges of the range.
Using $y+$r2a we get the index of the node to pass to add_output_node() on the first node at
$y+B<$r1a>.
And that's all there is to it!
For the fun of it, we'll take a quick look at the default connector.
Below is the actual default connector code, albeit a bit cleaned up, as well as
line numbers added.
= line 1 = sub _c {
= line 2 = my $self = shift;
= line 3 = my $r1a = shift;
= line 4 = my $r1b = shift;
= line 5 = my $r2a = shift;
= line 6 = my $r2b = shift;
= line 7 = my $mesh = $self->{mesh};
= line 8 = for my $y ($r1a..$r1b-1) {
= line 9 = for my $z ($r2a..$r2b-1) {
= line 10 = $mesh->[$y]->add_output_node($mesh->[$z]);
= line 11 = }
= line 12 = }
= line 12 = }
Its that easy! The simplest connector (well almost anyways). It just connects each
node in the first layer defined by ($r1a..$r1b) to every node in the second layer as
defined by ($r2a..$r2b).
Those of you that are still reading, if you do come up with any new connection functions,
PLEASE SEND THEM TO ME. I would love to see what others are doing, as well as get new
network ideas. I will probably include any connectors you send over in future releases (with
propoer credit and permission, of course).
Anyways, happy coding!
=head1 WHAT CAN IT DO?
Rodin Porrata asked on the ai-neuralnet-backprop malining list,
"What can they [Neural Networks] do?". In regards to that questioin,
consider the following:
Neural Nets are formed by simulated neurons connected together much the same
way the brain's neurons are, neural networks are able to associate and
generalize without rules. They have solved problems in pattern recognition,
robotics, speech processing, financial predicting and signal processing, to
name a few.
One of the first impressive neural networks was NetTalk, which read in ASCII
text and correctly pronounced the words (producing phonemes which drove a
speech chip), even those it had never seen before. Designed by John Hopkins
biophysicist Terry Sejnowski and Charles Rosenberg of Princeton in 1986,
this application made the Backprogagation training algorithm famous. Using
the same paradigm, a neural network has been trained to classify sonar
returns from an undersea mine and rock. This classifier, designed by
Sejnowski and R. Paul Gorman, performed better than a nearest-neighbor
classifier.
The kinds of problems best solved by neural networks are those that people
are good at such as association, evaluation and pattern recognition.
Problems that are difficult to compute and do not require perfect answers,
just very good answers, are also best done with neural networks. A quick,
very good response is often more desirable than a more accurate answer which
loan analysis and financial forecasting make good applications. New network
designers are working on weather forecasts by neural networks (Myself
included). Currently, doctors are developing medical neural networks as an
aid in diagnosis. Attorneys and insurance companies are also working on
neural networks to help estimate the value of claims.
Neural networks are poor at precise calculations and serial processing. They
are also unable to predict or recognize anything that does not inherently
contain some sort of pattern. For example, they cannot predict the lottery,
since this is a random process. It is unlikely that a neural network could
be built which has the capacity to think as well as a person does for two
reasons. Neural networks are terrible at deduction, or logical thinking and
the human brain is just too complex to completely simulate. Also, some
problems are too difficult for present technology. Real vision, for
example, is a long way off.
In short, Neural Networks are poor at precise calculations, but good at
association, evaluation, and pattern recognition.
=head1 EXAMPLES
Included are several example files in the "examples" directory from the
distribution ZIP file. Each of the examples includes a short explanation
at the top of the file. Each of these are ment to demonstrate simple, yet
practical (for the most part :-) uses of this module.
=head1 OTHER INCLUDED PACKAGES
These packages are not designed to be called directly, they are for internal use. They are
listed here simply for your refrence.
=item AI::NeuralNet::Mesh::node
This is the worker package of the mesh. It implements all the individual nodes of the mesh.
It might be good to look at the source for this package (in the Mesh.pm file) if you
plan to do a lot of or extensive custom node activation types.
=item AI::NeuralNet::Mesh::cap
This is applied to the input layer of the mesh to prevent the mesh from trying to recursivly
adjust weights out throug the inputs.
=item AI::NeuralNet::Mesh::output
This is simply a data collector package clamped onto the output layer to record the data
as it comes out of the mesh.
=head1 BUGS
This is a beta release of C<AI::NeuralNet::Mesh>, and that holding true, I am sure
there are probably bugs in here which I just have not found yet. If you find bugs in this module, I would
appreciate it greatly if you could report them to me at F<E<lt>jdb@wcoil.comE<gt>>,
or, even better, try to patch them yourself and figure out why the bug is being buggy, and
send me the patched code, again at F<E<lt>jdb@wcoil.comE<gt>>.
=head1 AUTHOR
Josiah Bryan F<E<lt>jdb@wcoil.comE<gt>>
Copyright (c) 2000 Josiah Bryan. All rights reserved. This program is free software;
you can redistribute it and/or modify it under the same terms as Perl itself.
The C<AI::NeuralNet::Mesh> and related modules are free software. THEY COME WITHOUT WARRANTY OF ANY KIND.
$Id: AI::NeuralNet::Mesh.pm, v0.44 2000/15/09 03:29:08 josiah Exp $
=head1 THANKS
Below are a list of the people that have contributed in some way to this module (no particular order):
Rodin Porrata, rodin@ursa.llnl.gov
Randal L. Schwartz, merlyn@stonehedge.com
Michiel de Roo, michiel@geo.uu.nl
Thanks to Randal and Michiel for spoting some documentation and makefile bugs in the last release.
Thanks to Rodin for continual suggetions and questions about the module and more.
=head1 DOWNLOAD
You can always download the latest copy of AI::NeuralNet::Mesh
from http://www.josiah.countystart.com/modules/get.pl?mesh:pod
=head1 MAILING LIST
A mailing list has been setup for AI::NeuralNet::Mesh and AI::NeuralNet::BackProp.
The list is for discussion of AI and neural net related topics as they pertain to
AI::NeuralNet::BackProp and AI::NeuralNet::mesh. I will also announce in the group
each time a new release of AI::NeuralNet::Mesh is available.
The list address is at:
ai-neuralnet-backprop@egroups.com
To subscribe, send a blank email:
ai-neuralnet-backprop-subscribe@egroups.com
=cut
AI-NeuralNet-SOM
view release on metacpan or search on metacpan
examples/eigenvector_initialization.pl view on Meta::CPAN
my $epsilon = 0.001;
my $epochs = 400;
{ # random initialisation
my $nn = new AI::NeuralNet::SOM::Rect (output_dim => "5x6",
input_dim => $dim);
$nn->initialize; # random
my @mes = $nn->train ($epochs, @vs);
warn "random: length until error is < $epsilon ". scalar (grep { $_ >= $epsilon } @mes);
}
{ # constant initialisation
my $nn = new AI::NeuralNet::SOM::Rect (output_dim => "5x6",
input_dim => $dim);
$nn->initialize ($vs[-1]);
my @mes = $nn->train ($epochs, @vs);
warn "constant: length until error is < $epsilon ". scalar (grep { $_ >= $epsilon } @mes);
}
{ # eigenvector initialisation
my $nn = new AI::NeuralNet::SOM::Rect (output_dim => "5x6",
input_dim => $dim);
my @training_vectors; # find these training vectors
{ # and prime them with this eigenvector stuff;
use PDL;
my $A = pdl \@vs;
while ($A->getdim(0) < $A->getdim(1)) { # make the beast quadratic
$A = append ($A, zeroes (1)); # by padding zeroes
}
my ($E, $e) = eigens_sym $A;
# print $E;
# print $e;
my @es = list $e; # eigenvalues
# warn "es : ".Dumper \@es;
my @es_desc = sort { $b <=> $a } @es; # eigenvalues sorted desc
# warn "desc: ".Dumper \@es_desc;
my @es_idx = map { _find_num ($_, \@es) } @es_desc; # eigenvalue indices sorted by eigenvalue (desc)
# warn "idx: ".Dumper \@es_idx;
sub _find_num {
my $v = shift;
my $l = shift;
for my $i (0..$#$l) {
return $i if $v == $l->[$i];
}
return undef;
}
for (@es_idx) { # from the highest values downwards, take the index
push @training_vectors, [ list $E->dice($_) ] ; # get the corresponding vector
}
}
$nn->initialize (@training_vectors[0..0]); # take only the biggest ones (the eigenvalues are big, actually)
#warn $nn->as_string;
my @mes = $nn->train ($epochs, @vs);
warn "eigen: length until error is < $epsilon ". scalar (grep { $_ >= $epsilon } @mes);
}
__END__
AI-NeuralNet-Simple
view release on metacpan or search on metacpan
examples/game_ai.pl view on Meta::CPAN
my $net = AI::NeuralNet::Simple->new(4,20,4);
$net->iterations(shift || 100000);
$net->train_set( [
# health knife gun enemy
[GOOD, YES, YES, 0], WANDER,
[GOOD, YES, NO, 2], HIDE,
[GOOD, YES, NO, 1], ATTACK,
[GOOD, YES, NO, 0], WANDER,
[GOOD, NO, YES, 2], ATTACK,
[GOOD, NO, YES, 1], ATTACK,
[GOOD, NO, NO, 3], HIDE,
[GOOD, NO, NO, 2], HIDE,
[GOOD, NO, NO, 1], RUN,
[GOOD, NO, NO, 0], WANDER,
[AVERAGE, YES, YES, 0], WANDER,
[AVERAGE, YES, NO, 2], HIDE,
[AVERAGE, YES, NO, 1], RUN,
[AVERAGE, NO, YES, 2], HIDE,
[AVERAGE, NO, YES, 1], ATTACK,
[AVERAGE, NO, NO, 3], HIDE,
[AVERAGE, NO, NO, 2], HIDE,
[AVERAGE, NO, NO, 1], RUN,
[AVERAGE, NO, NO, 0], WANDER,
[AVERAGE, NO, NO, 0], WANDER,
[POOR, YES, NO, 2], HIDE,
[POOR, YES, NO, 1], RUN,
[POOR, NO, YES, 2], HIDE,
[POOR, NO, YES, 1], RUN,
[POOR, NO, NO, 2], HIDE,
[POOR, NO, NO, 1], HIDE,
[POOR, NO, NO, 0], WANDER,
[POOR, YES, NO, 0], WANDER,
]);
my $format = "%8s %5s %3s %7s %6s\n";
my @actions = qw/attack run wander hide/;
examples/game_ai.pl view on Meta::CPAN
display_result($net,1,1,0,0);
display_result($net,1,0,1,2);
display_result($net,0,1,0,3);
while (1) {
print "Type 'quit' to exit\n";
my $health = prompt("Am I in poor, average, or good health? ", qr/^(?i:[pag])/);
my $knife = prompt("Do I have a knife? ", qr/^(?i:[yn])/);
my $gun = prompt("Do I have a gun? ", qr/^(?i:[yn])/);
my $enemies = prompt("How many enemies can I see? ", qr/^\d+$/);
$health = substr $health, 0, 1;
$health =~ tr/pag/012/;
foreach ($knife,$gun) {
$_ = substr $_, 0, 1;
tr/yn/10/;
}
printf "I think I will %s!\n\n", $actions[$net->winner([
$health,
$knife,
$gun,
$enemies])];
}
sub prompt
{
my ($message,$domain) = @_;
my $valid_response = 0;
my $response;
do {
print $message;
chomp($response = <STDIN>);
exit if substr(lc $response, 0, 1) eq 'q';
$valid_response = $response =~ /$domain/;
} until $valid_response;
return $response;
}
sub display_result
{
my ($net,@data) = @_;
my $result = $net->winner(\@data);
my @health = qw/Poor Average Good/;
my @knife = qw/No Yes/;
my @gun = qw/No Yes/;
printf $format,
$health[$_[1]],
$knife[$_[2]],
$gun[$_[3]],
$_[4], # number of enemies
$actions[$result];
}
AI-Ollama-Client
view release on metacpan or search on metacpan
lib/AI/Ollama/Client.pm view on Meta::CPAN
AI::Ollama::Client - Client for AI::Ollama
=head1 SYNOPSIS
use 5.020;
use AI::Ollama::Client;
my $client = AI::Ollama::Client->new(
server => 'https://example.com/',
);
my $res = $client->someMethod()->get;
say $res;
=head1 METHODS
=head2 C<< checkBlob >>
my $res = $client->checkBlob()->get;
Check to see if a blob exists on the Ollama server which is useful when creating models.
=cut
around 'checkBlob' => sub ( $super, $self, %options ) {
$super->( $self, %options )->then( sub( $res ) {
if( $res->code =~ /^2\d\d$/ ) {
return Future->done( 1 )
} else {
return Future->done( 0 )
}
});
};
=head2 C<< createBlob >>
my $res = $client->createBlob()->get;
Create a blob from a file. Returns the server file path.
=cut
=head2 C<< generateChatCompletion >>
my $res = $client->generateChatCompletion()->get;
Generate the next message in a chat with a provided model.
Returns a L<< AI::Ollama::GenerateChatCompletionResponse >>.
=cut
=head2 C<< copyModel >>
my $res = $client->copyModel()->get;
Creates a model with another name from an existing model.
=cut
=head2 C<< createModel >>
my $res = $client->createModel()->get;
Create a model from a Modelfile.
Returns a L<< AI::Ollama::CreateModelResponse >>.
=cut
=head2 C<< deleteModel >>
my $res = $client->deleteModel()->get;
Delete a model and its data.
=cut
=head2 C<< generateEmbedding >>
my $res = $client->generateEmbedding()->get;
Generate embeddings from a model.
Returns a L<< AI::Ollama::GenerateEmbeddingResponse >>.
=cut
=head2 C<< generateCompletion >>
use Future::Utils 'repeat';
my $responses = $client->generateCompletion();
repeat {
my ($res) = $responses->shift;
if( $res ) {
my $str = $res->get;
say $str;
}
Future::Mojo->done( defined $res );
} until => sub($done) { $done->get };
Generate a response for a given prompt with a provided model.
Returns a L<< AI::Ollama::GenerateCompletionResponse >>.
=cut
around 'generateCompletion' => sub ( $super, $self, %options ) {
# Encode images as base64, if images exist:
# (but create a copy so we don't over write the input array)
if (my $images = $options{images}) {
# Allow { filename => '/etc/passwd' }
$options{images} = [
map {
my $item = $_;
if( ref($item) eq 'HASH' ) {
$item = Mojo::File->new($item->{filename})->slurp();
};
encode_base64($item)
} @$images ];
}
return $super->($self, %options);
};
=head2 C<< pullModel >>
my $res = $client->pullModel(
name => 'llama',
)->get;
Download a model from the ollama library.
Returns a L<< AI::Ollama::PullModelResponse >>.
=cut
=head2 C<< pushModel >>
my $res = $client->pushModel()->get;
Upload a model to a model library.
Returns a L<< AI::Ollama::PushModelResponse >>.
=cut
=head2 C<< showModelInfo >>
my $info = $client->showModelInfo()->get;
say $info->modelfile;
Show details about a model including modelfile, template, parameters, license, and system prompt.
Returns a L<< AI::Ollama::ModelInfo >>.
=cut
=head2 C<< listModels >>
my $info = $client->listModels()->get;
for my $model ($info->models->@*) {
say $model->model; # llama2:latest
}
List models that are available locally.
Returns a L<< AI::Ollama::ModelsResponse >>.
AI-PBDD
view release on metacpan or search on metacpan
lib/AI/PBDD.pm view on Meta::CPAN
@ISA = qw(Exporter DynaLoader);
@EXPORT = qw(
BDD_REORDER_NONE
BDD_REORDER_WIN2
BDD_REORDER_WIN3
BDD_REORDER_SIFT
BDD_REORDER_RANDOM
);
@EXPORT_OK = qw(
// setup and cleanup
init
gc
verbose
kill
// simple BDD operations
getOne
getZero
createBDD
getVarCount
getBDD
// ref counting
ref
deref
localDeref
// BDD operations
and
or
andTo
orTo
nand
nor
xor
ite
imp
biimp
not
makeSet
exists
forall
relProd
restrict
constrain
// variables replacement
createPair
deletePair
replace
showPair
// BDD analysis
support
nodeCount
satOne
satCount
// printing
printDot
printSet
print
// debugging
printStats
checkPackage
debugPackage
debugBDD
// low-level access
internal_index
internal_refcount
internal_isconst
internal_constvalue
internal_iscomplemented
internal_then
internal_else
// dynamic variable ordering
reorder_setMethod
reorder_now
reorder_enableDynamic
reorder_createVariableGroup
);
$VERSION = '0.01';
bootstrap AI::PBDD $VERSION;
sub satCount {
my ($bdd, $vars_ignored) = @_;
if (!defined($vars_ignored)) {
return satCount__I($bdd);
} else {
return satCount__II($bdd, $vars_ignored);
}
}
sub printDot {
my ($bdd, $filename) = @_;
if (!defined($filename)) {
printDot__I($bdd);
} else {
printDot__II($bdd, $filename);
}
}
sub makeSet {
my ($vars, $size, $offset) = @_;
if (!defined($offset)) {
return makeSetI($vars, $size);
} else {
return makeSetII($vars, $size, $offset);
}
}
sub createPair {
my ($old, $new) = @_;
my $size = @$old;
return createPairI($old, $new, $size);
}
1;
__END__
lib/AI/PBDD.pm view on Meta::CPAN
Perl wrapper for the BuDDy C library
=head1 SYNOPSIS
use AI::PBDD qw(init createBDD and printDot kill);
init(100, 100000);
my $var1 = createBDD();
my $var2 = createBDD();
my $bdd = and($var1, $var2);
printDot($bdd);
kill();
=head1 DESCRIPTION
Binary Decision Diagrams (BDDs) are used for efficient computation of many common problems. This is done by giving a compact representation and a set of efficient operations on boolean functions f: {0,1}^n --> {0,1}.
lib/AI/PBDD.pm view on Meta::CPAN
BDDs and their operations are described in many academic papers that can be found on the Internet. A good place to get started with BDDs is the wikipedia article L<http://en.wikipedia.org/wiki/Binary_decision_diagram>.
It can also be useful to look at the test code for this package in the C<t> directory, as well as at the JBDD documentation and exaples at L<http://javaddlib.sourceforge.net/jbdd/>.
=head1 VERSION
This man page documents "PBDD" version 0.01.
=head1 AUTHOR
Gianluca Torta
mailto:torta@di.unito.it
=head1 COPYRIGHT
Copyright (c) 2011 by Gianluca Torta. All rights reserved.
AI-PSO
view release on metacpan or search on metacpan
examples/NeuralNet/pso_ann.pl view on Meta::CPAN
use strict;
use AI::PSO;
my %test_params = (
numParticles => 4,
numNeighbors => 3,
maxIterations => 1000,
dimensions => 8, # 8 since we have 8 network weights we want to optimize for a 3 input 2 hidden 1 output feed-forward neural net
deltaMin => -2.0,
deltaMax => 4.0,
meWeight => 2.0,
meMin => 0.0,
meMax => 1.0,
themWeight => 2.0,
themMin => 0.0,
themMax => 1.0,
exitFitness => 0.99,
verbose => 1,
);
my $numInputs = 3;
my $numHidden = 2;
my $xferFunc = "Logistic";
examples/NeuralNet/pso_ann.pl view on Meta::CPAN
my $expectedValue = 3.5; # this is the value that we want to train the ANN to produce (just like the example in t/PTO.t)
sub test_fitness_function(@) {
my (@arr) = (@_);
&writeAnnConfig($annConfig, $numInputs, $numHidden, $xferFunc, @arr);
my $netValue = &runANN($annConfig, $annInputs);
print "network value = $netValue\n";
# the closer the network value gets to our desired value
# then we want to set the fitness closer to 1.
#
# This is a special case of the sigmoid, and looks an awful lot
# like the hyperbolic tangent ;)
#
my $magnitudeFromBest = abs($expectedValue - $netValue);
return 2 / (1 + exp($magnitudeFromBest));
}
pso_set_params(\%test_params);
pso_register_fitness_function('test_fitness_function');
pso_optimize();
examples/NeuralNet/pso_ann.pl view on Meta::CPAN
##### io #########
sub writeAnnConfig() {
my ($configFile, $inputs, $hidden, $func, @weights) = (@_);
open(ANN, ">$configFile");
print ANN "$inputs $hidden\n";
print ANN "$func\n";
foreach my $weight (@weights) {
print ANN "$weight ";
}
print ANN "\n";
close(ANN);
}
sub runANN($$) {
my ($configFile, $dataFile) = @_;
my $networkValue = `ann_compute $configFile $dataFile`;
chomp($networkValue);
return $networkValue;
}
AI-ParticleSwarmOptimization-MCE
view release on metacpan or search on metacpan
example/PSOTest-MultiCore.pl view on Meta::CPAN
use AI::ParticleSwarmOptimization::MCE;
#use AI::ParticleSwarmOptimization::Pmap;
use Data::Dumper; $::Data::Dumper::Sortkeys = 1;
#=======================================================================
sub calcFit {
my @values = @_;
my $offset = int (-@values / 2);
my $sum;
select( undef, undef, undef, 0.01 ); # Simulation of heavy processing...
$sum += ($_ - $offset++) ** 2 for @values;
return $sum;
}
#=======================================================================
++$|;
#-----------------------------------------------------------------------
#my $pso = AI::ParticleSwarmOptimization::Pmap->new( # Multi-core
my $pso = AI::ParticleSwarmOptimization::MCE->new( # Multi-core
#my $pso = AI::ParticleSwarmOptimization->new( # Single-core
-fitFunc => \&calcFit,
-dimensions => 10,
-iterations => 10,
-numParticles => 1000,
# only for many-core version # the best if == $#cores of your system
# selecting best value if undefined
-workers => 4,
);
my $beg = time;
AI-ParticleSwarmOptimization-Pmap
view release on metacpan or search on metacpan
example/PSOTest-MultiCore.pl view on Meta::CPAN
#use AI::ParticleSwarmOptimization::MCE;
use AI::ParticleSwarmOptimization::Pmap;
use Data::Dumper; $::Data::Dumper::Sortkeys = 1;
#=======================================================================
sub calcFit {
my @values = @_;
my $offset = int (-@values / 2);
my $sum;
select( undef, undef, undef, 0.01 ); # Simulation of heavy processing...
$sum += ($_ - $offset++) ** 2 for @values;
return $sum;
}
#=======================================================================
++$|;
#-----------------------------------------------------------------------
#my $pso = AI::ParticleSwarmOptimization->new( # Single-core
#my $pso = AI::ParticleSwarmOptimization::MCE->new( # Multi-core
my $pso = AI::ParticleSwarmOptimization::Pmap->new( # Multi-core
-fitFunc => \&calcFit,
-dimensions => 10,
-iterations => 10,
-numParticles => 1000,
# only for many-core version # the best if == $#cores of your system
# selecting best value if undefined
-workers => 4,
);
my $beg = time;
AI-ParticleSwarmOptimization
view release on metacpan or search on metacpan
Samples/PSOPlatTest.pl view on Meta::CPAN
use AI::ParticleSwarmOptimization;
use Math::Random::MT qw();
++$|;
my $pso = AI::ParticleSwarmOptimization->new (
-fitFunc => \&calcFit,
-dimensions => 3,
-iterations => 500,
-exitPlateau => 1,
-exitPlateauDP => 3,
-exitPlateauBurnin => 100,
-exitPlateauWindow => 60,
);
$pso->init ();
my $fitValue = $pso->optimize ();
Samples/PSOPlatTest.pl view on Meta::CPAN
my ($fit, @values) = $pso->getParticleBestPos ($best);
my $iters = $pso->getIterationCount ();
print $pso->getSeed();
printf ",# Fit %.5f at (%s) after %d iterations\n",
$fit, join (', ', map {sprintf '%.4f', $_} @values), $iters;
sub calcFit {
my @values = @_;
my $offset = int (-@values / 2);
my $sum;
$sum += ($_ - $offset++)**2 for @values;
return $sum;
}
AI-Pathfinding-AStar-Rectangle
view release on metacpan or search on metacpan
Benchmark/perl-vs-xs.pl view on Meta::CPAN
use AI::Pathfinding::AStar::Rectangle;
my $m = AI::Pathfinding::AStar::Rectangle->new({ width => WIDTH_X, heigth => WIDTH_Y });
for my $x (0 .. WIDTH_X - 1 )
{
for my $y (0 .. WIDTH_Y - 1 )
{
$map[$x][$y] = 1;
}
}
$map[5][$_] = 0 for 5 .. WIDTH_Y - 5;
$map[WIDTH_X - 5][$_] = 0 for 5 .. WIDTH_Y - 5;
$map[$_][5] = 0 for 5 .. WIDTH_X - 5;
Benchmark/perl-vs-xs.pl view on Meta::CPAN
$map[WIDTH_X - 15][$_] = 0 for 15 .. WIDTH_Y - 10;
$map[$_][15] = 0 for 15 .. WIDTH_X - 15;
for my $x (0 .. WIDTH_X - 1 )
{
for my $y (0 .. WIDTH_Y - 1 )
{
$m->set_passability($x, $y, $map[$x][$y]) ;
}
}
my ( $x_start, $y_start ) = ( WIDTH_X >> 1, WIDTH_Y >> 1 );
my ( $x_end, $y_end ) = ( 0, 0 );
my $t0 = [gettimeofday];
my $path;
my $r = timethese( -1, {Perl=>sub { astar( $x_start, $y_start, $x_end, $y_end ) },
XS=>sub {$m->astar($x_start, $y_start, $x_end, $y_end);}});
cmpthese($r);
die;
for (0..99) {
$path = &astar( $x_start, $y_start, $x_end, $y_end );
}
print "Elapsed: ".tv_interval ( $t0 )."\n";
print "Path length: ".length($path)."\n";
# start end points
$map[ $x_start ][ $y_start ] = 3;
$map[ $x_end ][ $y_end ] = 4;
# draw path
my %vect = (
# x y
1 => [-1, 1, '|/'],
2 => [ 0, 1, '.|'],
3 => [ 1, 1, '|\\'],
4 => [-1, 0, '|<'],
6 => [ 1, 0, '|>'],
7 => [-1,-1, '|\\'],
8 => [ 0,-1, '\'|'],
9 => [ 1,-1, '|/']
);
my ( $x, $y ) = ( $x_start, $y_start );
for ( split //, $path )
{
$map[$x][$y] = '|o';
$x += $vect{$_}->[0];
$y += $vect{$_}->[1];
$map[$x][$y] = '|o';
}
printf "%02d", $_ for 0 .. WIDTH_X - 1;
print "\n";
for my $y ( 0 .. WIDTH_Y - 1 )
{
for my $x ( 0 .. WIDTH_X - 1 )
{
print $map[$x][$y] eq
'1' ? "|_" : (
$map[$x][$y] eq '0' ? "|#" : (
$map[$x][$y] eq '3' ? "|S" : (
$map[$x][$y] eq '4' ? "|E" : $map[$x][$y] ) ) );
}
print "$y\n";
}
sub astar
{
my ( $xs, $ys, $xe, $ye ) = @_;
my %close;
my ( %open, @g, @h, @r, @open_idx );
for my $x (0 .. WIDTH_X - 1 )
{
for my $y (0 .. WIDTH_Y - 1 )
{
$g[$x][$y] = 0;
$r[$x][$y] = 0;
$h[$x][$y] = 0;
}
}
my %cost = (
"0.-1" => 5, #|.
"1.-1" => 7, #/.
"1.0" => 5, #.-
"1.1" => 7, #`\
"0.1" => 5, #`|
"-1.1" => 7, #
"-1.0" => 5,
"-1.-1" => 7
);
my $it = 0;
my $oindx = 0;
my ( $x, $y ) = ( $xs, $ys );
while ( $x != $xe || $y != $ye )
{
$close{$x}{$y} = 1;
$open{$x}{$y} = 0;
for ( "0.-1", "-1.1", "0.1", "1.1", "-1.0", "1.-1", "1.0", "-1.-1" )
{
my ( $xd, $yd ) = split /\./, $_;
my ( $xn, $yn ) = ( $x + $xd, $y + $yd );
next if $xn == WIDTH_X ||
$xn < 0 ||
$yn == WIDTH_Y ||
$yn < 0 ||
$close{$xn}{$yn} ||
$map[$xn][$yn] == 0;
my $ng = $g[$x][$y] + $cost{$_};
if ( $open{$xn}{$yn} )
{
if ( $ng < $g[$xn][$yn] )
{
$r[$xn][$yn] = [$x,$y];
$g[$xn][$yn] = $ng;
}
}
else
{
$open{$xn}{$yn} = 1;
$g[$xn][$yn] = $ng;
my ( $xa, $ya ) = ( abs( $xn - $xe ), abs( $yn - $ye ) );
$h[$xn][$yn] = #( $xa > $ya ? $xa : $ya ) * 7;
( abs( $xn - $xe ) + abs( $yn - $ye ) ) * 7;
$r[$xn][$yn] = [$x,$y];
push @open_idx, [$xn, $yn, \$g[$xn][$yn], \$h[$xn][$yn]];
}
# deb($x, $y, $xn, $yn, \@g);
}
@open_idx = sort { ${$a->[2]} + ${$a->[3]} <=> ${$b->[2]} + ${$b->[3]} } @open_idx;
( $x, $y ) = @{ shift @open_idx };
$it++;
}
# print "Iterations: $it: $oindx\n";
my $path = "";
my %idx2path =
(
"0.-1" => 8, #|.
"1.-1" => 9, #/.
"1.0" => 6, #.-
"1.1" => 3, #`\
"0.1" => 2, #`|
"-1.1" => 1, #
"-1.0" => 4,
"-1.-1" => 7
);
while ( $x != $xs || $y != $ys )
{
# print "$x:$y\n";
my ($xp, $yp) = @{$r[$x][$y]};
$path = $idx2path{($x-$xp).".".($y-$yp)}.$path;
( $x, $y ) = ( $xp, $yp);
}
# print "Path: $path\n";
return $path;
}
sub calc_obstacle
{
my ( $x1, $y1, $x2, $y2 ) = @_;
my ( $x, $y, $Xend, $obstacle, $pixel);
my $dx = abs($x2 - $x1);
my $dy = abs($y2 - $y1);
my $d = ( $dy << 1 ) - $dx;
my $inc1 = $dy << 1;
my $inc2 = ($dy - $dx) << 1;
if ( $x1 > $x2)
{
$x = $x2;
$y = $y2;
$Xend = $x1;
}
else
{
$x = $x1;
$y = $y1;
$Xend = $x2;
};
$obstacle+=!$map[$x][$y];
$pixel+=5;
while ( $x < $Xend )
{
$x++;
if ($d < 0) {$d += $inc1}
else
{
$y++;
$d += $inc2;
};
$obstacle+=!$map[$x][$y];
$pixel += 5;
};
return ( $obstacle << 3 ) + $pixel;
}
sub deb
{
my ( $x, $y, $xn, $yn, $g) = @_;
for my $j ( 0 .. WIDTH_Y - 1 )
{
for my $i ( 0 .. WIDTH_X - 1 )
{
if ( !$map[$i][$j] )
{
print " ##"
}
else
{
if ( $x == $i && $y == $j)
{
print "c";
}
elsif ( $xn == $i && $yn == $j )
{
print "n";
}
else
{
print " ";
}
printf "%02d", $g->[$i]->[$j]
}
}
print "\n";
}
<>;
}
AI-Pathfinding-AStar
view release on metacpan or search on metacpan
lib/AI/Pathfinding/AStar.pm view on Meta::CPAN
use AI::Pathfinding::AStar::AStarNode;
my $nodes;
sub _init {
my $self = shift;
croak "no getSurrounding() method defined" unless $self->can("getSurrounding");
return $self->SUPER::_init(@_);
}
sub doAStar
{
my ($map, $target, $open, $nodes, $max) = @_;
my $n = 0;
FLOOP: while ( (defined $open->top()) && ($open->top()->{id} ne $target) ) {
#allow incremental calculation
last FLOOP if (defined($max) and (++$n == $max));
my $curr_node = $open->extract_top();
$curr_node->{inopen} = 0;
my $G = $curr_node->{g};
#get surrounding squares
my $surr_nodes = $map->getSurrounding($curr_node->{id}, $target);
foreach my $node (@$surr_nodes) {
my ($surr_id, $surr_cost, $surr_h) = @$node;
#skip the node if it's in the CLOSED list
next if ( (exists $nodes->{$surr_id}) && (! $nodes->{$surr_id}->{inopen}) );
#add it if we haven't seen it before
if (! exists $nodes->{$surr_id}) {
my $surr_node = AI::Pathfinding::AStar::AStarNode->new($surr_id,$G+$surr_cost,$surr_h);
$surr_node->{parent} = $curr_node;
$surr_node->{cost} = $surr_cost;
$surr_node->{inopen} = 1;
$nodes->{$surr_id} = $surr_node;
$open->add($surr_node);
}
else {
#otherwise it's already in the OPEN list
#check to see if it's cheaper to go through the current
#square compared to the previous path
my $surr_node = $nodes->{$surr_id};
my $currG = $surr_node->{g};
my $possG = $G + $surr_cost;
if ($possG < $currG) {
#change the parent
$surr_node->{parent} = $curr_node;
$surr_node->{g} = $possG;
$open->decrease_key($surr_node);
}
}
}
}
}
sub fillPath
{
my ($map,$open,$nodes,$target) = @_;
my $path = [];
my $curr_node = (exists $nodes->{$target}) ? $nodes->{$target} : $open->top();
while (defined $curr_node) {
unshift @$path, $curr_node->{id};
$curr_node = $curr_node->{parent};
}
return $path;
}
sub findPath {
my ($map, $start, $target) = @_;
my $nodes = {};
my $curr_node = undef;
my $open = Heap::Binomial->new;
#add starting square to the open list
$curr_node = AI::Pathfinding::AStar::AStarNode->new($start,0,0); # AStarNode(id,g,h)
$curr_node->{parent} = undef;
$curr_node->{cost} = 0;
$curr_node->{g} = 0;
$curr_node->{h} = 0;
$curr_node->{inopen} = 1;
$nodes->{$start} = $curr_node;
$open->add($curr_node);
$map->doAStar($target,$open,$nodes,undef);
my $path = $map->fillPath($open,$nodes,$target);
return wantarray ? @{$path} : $path;
}
sub findPathIncr {
my ($map, $start, $target, $state, $max) = @_;
my $open = undef;
my $curr_node = undef;;
my $nodes = {};
if (defined($state)) {
$nodes = $state->{'visited'};
$open = $state->{'open'};
}
else {
$open = Heap::Binomial->new;
#add starting square to the open list
$curr_node = AI::Pathfinding::AStar::AStarNode->new($start,0,0); # AStarNode(id,g,h)
$curr_node->{parent} = undef;
$curr_node->{cost} = 0;
$curr_node->{g} = 0;
$curr_node->{h} = 0;
$curr_node->{inopen} = 1;
$nodes->{$start} = $curr_node;
$open->add($curr_node);
}
$map->doAStar($target,$open,$nodes,$max);
my $path = $map->fillPath($open,$nodes,$target);
$state = {
'path' => $path,
'open' => $open,
'visited' => $nodes,
'done' => defined($nodes->{$target}),
};
return $state;
}
1;
__END__
lib/AI/Pathfinding/AStar.pm view on Meta::CPAN
AI::Pathfinding::AStar - Perl implementation of the A* pathfinding algorithm
=head1 SYNOPSIS
package My::Map::Package;
use base AI::Pathfinding::AStar;
# Methods required by AI::Pathfinding::AStar
sub getSurrounding { ... }
package main;
use My::Map::Package;
my $map = My::Map::Package->new or die "No map for you!";
my $path = $map->findPath($start, $target);
print join(', ', @$path), "\n";
#Or you can do it incrementally, say 3 nodes at a time
my $state = $map->findPathIncr($start, $target, undef, 3);
while ($state->{path}->[-1] ne $target) {
print join(', ', @{$state->{path}}), "\n";
$state = $map->findPathIncr($start, $target, $state, 3);
}
print "Completed Path: ", join(', ', @{$state->{path}}), "\n";
=head1 DESCRIPTION
This module implements the A* pathfinding algorithm. It acts as a base class from which a custom map object can be derived. It requires from the map object a subroutine named C<getSurrounding> (described below) and provides to the object two routin...
AI::Pathfinding::AStar requires that the map object define a routine named C<getSurrounding> which accepts the starting and target node ids for which you are calculating the path. In return it should provide an array reference containing the followi...
AI-Pathfinding-OptimizeMultiple
view release on metacpan or search on metacpan
lib/AI/Pathfinding/OptimizeMultiple.pm view on Meta::CPAN
has _num_boards => ( isa => 'Int', is => 'ro', init_arg => 'num_boards', );
has _orig_scans_data => ( isa => 'PDL', is => 'rw' );
has _optimize_for => ( isa => 'Str', is => 'ro', init_arg => 'optimize_for', );
has _scans_data => ( isa => 'PDL', is => 'rw' );
has _selected_scans =>
( isa => 'ArrayRef', is => 'ro', init_arg => 'selected_scans', );
has _status => ( isa => 'Str', is => 'rw' );
has _quotas => ( isa => 'ArrayRef[Int]', is => 'ro', init_arg => 'quotas' );
has _total_boards_solved => ( isa => 'Int', is => 'rw' );
has _total_iters => ( is => 'rw' );
has _trace_cb =>
( isa => 'Maybe[CodeRef]', is => 'ro', init_arg => 'trace_cb' );
has _scans_meta_data => ( isa => 'ArrayRef', is => 'ro', init_arg => 'scans' );
has _scans_iters_pdls =>
( isa => 'HashRef', is => 'rw', init_arg => 'scans_iters_pdls' );
has _stats_factors => (
isa => 'HashRef',
is => 'ro',
init_arg => 'stats_factors',
default => sub { return +{}; },
);
sub BUILD
{
my $self = shift;
my $args = shift;
my $scans_data = PDL::cat(
map {
my $id = $_->id();
my $pdl = $self->_scans_iters_pdls()->{$id};
my $factor = $self->_stats_factors->{$id};
(
defined($factor)
? ( ( $pdl >= 0 ) * ( ( $pdl / $factor )->ceil() ) +
( $pdl < 0 ) * $pdl )
: $pdl
);
} @{ $self->_selected_scans() }
);
$self->_orig_scans_data($scans_data);
$self->_scans_data( $self->_orig_scans_data()->copy() );
return 0;
}
my $BOARDS_DIM = 0;
my $SCANS_DIM = 1;
my $STATISTICS_DIM = 2;
sub _next_iter_idx
{
my $self = shift;
my $ret = $self->_iter_idx();
$self->_iter_idx( $ret + 1 );
return $ret;
}
sub _get_next_quota
{
my $self = shift;
my $iter = $self->_next_iter_idx();
if ( ref( $self->_quotas() ) eq "ARRAY" )
{
return $self->_quotas()->[$iter];
}
else
{
return $self->_quotas()->($iter);
}
}
sub _calc_get_iter_state_param_method
{
my $self = shift;
my $optimize_for = $self->_optimize_for();
my %resolve = (
len => "_get_iter_state_params_len",
minmax_len => "_get_iter_state_params_minmax_len",
speed => "_get_iter_state_params_speed",
);
return $resolve{$optimize_for};
}
sub _get_iter_state_params
{
my $self = shift;
my $method = $self->_calc_get_iter_state_param_method();
return $self->$method();
}
sub _my_sum_over
{
my $pdl = shift;
return $pdl->sumover()->slice(":,(0)");
}
sub _my_xchg_sum_over
{
my $pdl = shift;
return _my_sum_over( $pdl->xchg( 0, 1 ) );
}
sub _get_iter_state_params_len
{
my $self = shift;
my $iters_quota = 0;
my $num_solved_in_iter = 0;
my $selected_scan_idx;
# If no boards were solved, then try with a larger quota
while ( $num_solved_in_iter == 0 )
{
my $q_more = $self->_get_next_quota();
if ( !defined($q_more) )
{
AI::Pathfinding::OptimizeMultiple::Error::OutOfQuotas->throw(
error => "No q_more", );
}
$iters_quota += $q_more;
my $iters = $self->_scans_data()->slice(":,:,0");
my $solved = ( ( $iters <= $iters_quota ) & ( $iters > 0 ) );
my $num_moves = $self->_scans_data->slice(":,:,2");
my $solved_moves = $solved * $num_moves;
my $solved_moves_sums = _my_sum_over($solved_moves);
my $solved_moves_counts = _my_sum_over($solved);
my $solved_moves_avgs = $solved_moves_sums / $solved_moves_counts;
( undef, undef, $selected_scan_idx, undef ) =
$solved_moves_avgs->minmaximum();
$num_solved_in_iter = $solved_moves_counts->at($selected_scan_idx);
}
return {
quota => $iters_quota,
num_solved => $num_solved_in_iter,
scan_idx => $selected_scan_idx,
};
}
sub _get_iter_state_params_minmax_len
{
my $self = shift;
my $iters_quota = 0;
my $num_solved_in_iter = 0;
my $selected_scan_idx;
# If no boards were solved, then try with a larger quota
while ( $num_solved_in_iter == 0 )
{
my $q_more = $self->_get_next_quota();
if ( !defined($q_more) )
{
AI::Pathfinding::OptimizeMultiple::Error::OutOfQuotas->throw(
error => "No q_more", );
}
$iters_quota += $q_more;
my $iters = $self->_scans_data()->slice(":,:,0");
my $solved = ( ( $iters <= $iters_quota ) & ( $iters > 0 ) );
my $num_moves = $self->_scans_data->slice(":,:,2");
my $solved_moves = $solved * $num_moves;
my $solved_moves_maxima = $solved_moves->maximum()->slice(":,(0),(0)");
my $solved_moves_counts = _my_sum_over($solved);
( undef, undef, $selected_scan_idx, undef ) =
$solved_moves_maxima->minmaximum();
$num_solved_in_iter = $solved_moves_counts->at($selected_scan_idx);
}
return {
quota => $iters_quota,
num_solved => $num_solved_in_iter,
scan_idx => $selected_scan_idx,
};
}
sub _get_iter_state_params_speed
{
my $self = shift;
my $iters_quota = 0;
my $num_solved_in_iter = 0;
my $selected_scan_idx;
# If no boards were solved, then try with a larger quota
while ( $num_solved_in_iter == 0 )
{
my $q_more = $self->_get_next_quota();
if ( !defined($q_more) )
{
AI::Pathfinding::OptimizeMultiple::Error::OutOfQuotas->throw(
error => "No q_more" );
}
$iters_quota += $q_more;
( undef, $num_solved_in_iter, undef, $selected_scan_idx ) =
PDL::minmaximum(
PDL::sumover(
( $self->_scans_data() <= $iters_quota ) &
( $self->_scans_data() > 0 )
)
);
}
return {
quota => $iters_quota,
num_solved => $num_solved_in_iter->at(0),
scan_idx => $selected_scan_idx->at(0),
};
}
sub _get_selected_scan
{
my $self = shift;
my $iter_state =
AI::Pathfinding::OptimizeMultiple::IterState->new(
$self->_get_iter_state_params(), );
$iter_state->attach_to($self);
return $iter_state;
}
sub _inspect_quota
{
my $self = shift;
my $state = $self->_get_selected_scan();
$state->register_params();
$state->update_total_iters();
if ( $self->_total_boards_solved() == $self->_num_boards() )
{
$self->_status("solved_all");
}
else
{
$state->update_idx_slice();
}
$state->detach();
}
sub calc_meta_scan
{
my $self = shift;
$self->chosen_scans( [] );
$self->_total_boards_solved(0);
$self->_total_iters(0);
$self->_status("iterating");
# $self->_inspect_quota() throws ::Error::OutOfQuotas if
# it does not have any available quotas.
eval {
while ( $self->_status() eq "iterating" )
{
$self->_inspect_quota();
}
};
if (
my $err = Exception::Class->caught(
'AI::Pathfinding::OptimizeMultiple::Error::OutOfQuotas')
)
{
$self->_status("out_of_quotas");
}
else
{
$err = Exception::Class->caught();
if ($err)
{
if ( not( blessed $err && $err->can('rethrow') ) )
{
die $err;
}
$err->rethrow;
}
}
return;
}
sub _get_num_scans
{
my $self = shift;
return ( ( $self->_scans_data()->dims() )[$SCANS_DIM] );
}
sub _calc_chosen_scan
{
my ( $self, $selected_scan_idx, $iters_quota ) = @_;
return AI::Pathfinding::OptimizeMultiple::ScanRun->new(
{
iters => (
$iters_quota * (
$self->_stats_factors->{
( $self->_selected_scans->[$selected_scan_idx]->id() ),
} // 1
)
),
scan_idx => $selected_scan_idx,
}
);
}
sub calc_flares_meta_scan
{
my $self = shift;
$self->chosen_scans( [] );
$self->_total_boards_solved(0);
$self->_total_iters(0);
$self->_status("iterating");
my $iters_quota = 0;
my $flares_num_iters = PDL::Core::pdl( [ (0) x $self->_get_num_scans() ] );
my $ones_constant =
PDL::Core::pdl( [ map { [1] } ( 1 .. $self->_get_num_scans() ) ] );
my $next_num_iters_for_each_scan_x_scan =
( ( $ones_constant x $flares_num_iters ) );
my $num_moves = $self->_scans_data->slice(":,:,1");
# The number of moves for dimension 0,1,2 above.
my $num_moves_repeat = $num_moves->clump( 1 .. 2 )->xchg( 0, 1 )
->dummy( 0, $self->_get_num_scans() );
my $selected_scan_idx;
my $loop_iter_num = 0;
my $UNSOLVED_NUM_MOVES_CONSTANT = 64 * 1024 * 1024;
my $last_avg = $UNSOLVED_NUM_MOVES_CONSTANT;
FLARES_LOOP:
while ( my $q_more = $self->_get_next_quota() )
{
$iters_quota += $q_more;
# Next number of iterations for each scan x scan combination.
my $next_num_iters = (
( $ones_constant x $flares_num_iters ) + (
PDL::MatrixOps::identity( $self->_get_num_scans() ) *
$iters_quota
)
);
# print "\$next_num_iters = $next_num_iters\n";
my $iters = $self->_scans_data()->slice(":,:,0");
my $iters_repeat =
$iters->dummy( 0, $self->_get_num_scans() )->xchg( 1, 2 )
->clump( 2 .. 3 );
# print "\$iters_repeat =", join(",",$iters_repeat->dims()), "\n";
my $next_num_iters_repeat =
$next_num_iters->dummy( 0, $self->_num_boards() )->xchg( 0, 2 );
# print "\$next_num_iters_repeat =", join(",",$next_num_iters_repeat->dims()), "\n";
# A boolean tensor of which boards were solved:
# Dimension 0 - Which scan is it. - size - _get_num_scans()
# Dimension 1 - Which scan we added the quota to
# - size - _get_num_scans()
# Dimension 2 - Which board. - size - _num_boards()
my $solved =
( $iters_repeat >= 0 ) * ( $iters_repeat < $next_num_iters_repeat );
# print "\$num_moves_repeat =", join(",",$num_moves_repeat->dims()), "\n";
my $num_moves_solved =
( $solved * $num_moves_repeat ) +
( $solved->not() * $UNSOLVED_NUM_MOVES_CONSTANT );
my $minimal_num_moves_solved =
$num_moves_solved->xchg( 0, 1 )->minimum();
my $which_minima_are_solved =
( $minimal_num_moves_solved != $UNSOLVED_NUM_MOVES_CONSTANT );
my $minimal_with_zeroes =
$which_minima_are_solved * $minimal_num_moves_solved;
my $solved_moves_sums = _my_xchg_sum_over($minimal_with_zeroes);
my $solved_moves_counts = _my_xchg_sum_over($which_minima_are_solved);
my $solved_moves_avgs = $solved_moves_sums / $solved_moves_counts;
# print join(",", $solved_moves_avgs->minmaximum()), "\n";
my $min_avg;
( $min_avg, undef, $selected_scan_idx, undef ) =
$solved_moves_avgs->minmaximum();
$last_avg = $min_avg;
push @{ $self->chosen_scans() },
$self->_calc_chosen_scan( $selected_scan_idx, $iters_quota );
$flares_num_iters->set( $selected_scan_idx,
$flares_num_iters->at($selected_scan_idx) + $iters_quota );
$self->_selected_scans()->[$selected_scan_idx]->mark_as_used();
$iters_quota = 0;
my $num_solved = $solved_moves_counts->at($selected_scan_idx);
my $flares_num_iters_repeat =
$flares_num_iters->dummy( 0, $self->_num_boards() );
# A boolean tensor:
# Dimension 0 - board.
# Dimension 1 - scans.
my $solved_with_which_iter =
( $flares_num_iters_repeat >= $iters->clump( 1 .. 2 ) ) &
( $iters->clump( 1 .. 2 ) >= 0 );
my $total_num_iters = (
( $solved_with_which_iter * $flares_num_iters_repeat )->sum() + (
$solved_with_which_iter->not()->andover() *
$flares_num_iters->sum()
)->sum()
);
print "Finished ", $loop_iter_num++,
" ; #Solved = $num_solved ; Iters = $total_num_iters ; Avg = $min_avg\n";
STDOUT->flush();
}
}
sub calc_board_iters
{
my $self = shift;
my $board = shift;
my $board_iters = 0;
my @info = PDL::list( $self->_orig_scans_data()->slice("$board,:") );
my @orig_info = @info;
foreach my $s ( @{ $self->chosen_scans() } )
{
if ( ( $info[ $s->scan_idx() ] > 0 )
&& ( $info[ $s->scan_idx() ] <= $s->iters() ) )
{
$board_iters += $info[ $s->iters() ];
last;
}
else
{
if ( $info[ $s->scan_idx() ] > 0 )
{
$info[ $s->scan_idx() ] -= $s->iters();
}
$board_iters += $s->iters();
}
}
return {
'per_scan_iters' => \@orig_info,
'board_iters' => $board_iters,
};
}
sub get_final_status
{
my $self = shift;
return $self->_status();
}
sub simulate_board
{
my ( $self, $board_idx, $args ) = @_;
if ( $board_idx !~ /\A[0-9]+\z/ )
{
die "Board index '$board_idx' is not numeric!";
}
$args ||= {};
my $chosen_scans = ( $args->{chosen_scans} || $self->chosen_scans );
my @info = PDL::list( $self->_orig_scans_data()->slice("$board_idx,:") );
my $board_iters = 0;
my @scan_runs;
my $status = "Unsolved";
my $add_new_scan_run = sub {
my $scan_run = shift;
push @scan_runs, $scan_run;
$board_iters += $scan_run->iters();
return;
};
SCANS_LOOP:
foreach my $s (@$chosen_scans)
{
if ( ( $info[ $s->scan_idx() ] > 0 )
&& ( $info[ $s->scan_idx() ] <= $s->iters() ) )
{
$add_new_scan_run->(
AI::Pathfinding::OptimizeMultiple::ScanRun->new(
{
iters => $info[ $s->scan_idx() ],
scan_idx => $s->scan_idx(),
},
)
);
$status = "Solved";
last SCANS_LOOP;
}
else
{
if ( $info[ $s->scan_idx() ] > 0 )
{
$info[ $s->scan_idx() ] -= $s->iters();
}
$add_new_scan_run->(
AI::Pathfinding::OptimizeMultiple::ScanRun->new(
{
iters => $s->iters(),
scan_idx => $s->scan_idx(),
},
)
);
}
}
return AI::Pathfinding::OptimizeMultiple::SimulationResults->new(
{
status => $status,
scan_runs => \@scan_runs,
total_iters => $board_iters,
}
);
}
sub _trace
{
my ( $self, $args ) = @_;
if ( my $trace_callback = $self->_trace_cb() )
{
$trace_callback->($args);
}
return;
}
sub get_total_iters
{
my $self = shift;
return $self->_total_iters();
}
sub _add_to_total_iters
{
my $self = shift;
my $how_much = shift;
$self->_total_iters( $self->_total_iters() + $how_much );
return;
}
sub _add_to_total_boards_solved
{
my $self = shift;
my $how_much = shift;
$self->_total_boards_solved( $self->_total_boards_solved() + $how_much );
return;
}
1; # End of AI::Pathfinding::OptimizeMultiple
__END__
lib/AI/Pathfinding/OptimizeMultiple.pm view on Meta::CPAN
version 0.0.17
=head1 SYNOPSIS
use AI::Pathfinding::OptimizeMultiple
my @scans =
(
{
name => "first_search"
},
{
name => "second_search",
},
{
name => "third_search",
},
);
my $obj = AI::Pathfinding::OptimizeMultiple->new(
{
scans => \@scans,
num_boards => 32_000,
optimize_for => 'speed',
scans_iters_pdls =>
{
first_search => $first_search_pdl,
second_search => $second_search_pdl,
},
quotas => [400, 300, 200],
selected_scans =>
[
AI::Pathfinding::OptimizeMultiple::Scan->new(
id => 'first_search',
cmd_line => "--preset first_search",
),
AI::Pathfinding::OptimizeMultiple::Scan->new(
id => 'second_search',
cmd_line => "--preset second_search",
),
AI::Pathfinding::OptimizeMultiple::Scan->new(
id => 'third_search',
cmd_line => "--preset third_search",
),
],
}
);
$obj->calc_meta_scan();
foreach my $scan_alloc (@{$self->chosen_scans()})
{
printf "Run %s for %d iterations.\n",
$scans[$scan_alloc->scan_idx], $scan_alloc->iters;
}
=head1 DESCRIPTION
This CPAN distribution implements the algorithm described here:
lib/AI/Pathfinding/OptimizeMultiple.pm view on Meta::CPAN
with it, or whatever. If you want to contribute patches, please send me a diff or prod me to pull
from your repository :)
L<http://github.com/shlomif/fc-solve>
git clone ssh://git@github.com/shlomif/fc-solve.git
=head1 AUTHOR
Shlomi Fish <shlomif@cpan.org>
lib/AI/Pathfinding/OptimizeMultiple.pm view on Meta::CPAN
This software is Copyright (c) 2012 by Shlomi Fish.
This is free software, licensed under:
The MIT (X11) License
=cut
AI-Pathfinding-SMAstar
view release on metacpan or search on metacpan
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
# This allows declaration use AI::Pathfinding::SMAstar ':all';
# If you do not need this, moving things directly into @EXPORT or @EXPORT_OK
# will save memory.
our %EXPORT_TAGS = ( 'all' => [ qw(
) ] );
our @EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } );
our @EXPORT = qw(
);
our $VERSION = '0.07';
use AI::Pathfinding::SMAstar::PriorityQueue;
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
##################################################
# SMAstar constructor
##################################################
sub new {
my $invocant = shift;
my $class = ref($invocant) || $invocant;
my $self = {
_priority_queue => AI::Pathfinding::SMAstar::PriorityQueue->new(),
_state_eval_func => undef,
_state_goal_p_func => undef,
_state_num_successors_func => undef,
_state_successors_iterator => undef,
_show_prog_func => undef,
_state_get_data_func => undef,
@_, # attribute override
};
return bless $self, $class;
}
sub state_eval_func {
my $self = shift;
if (@_) { $self->{_state_eval_func} = shift }
return $self->{_state_eval_func};
}
sub state_goal_p_func {
my $self = shift;
if (@_) { $self->{_state_goal_p_func} = shift }
return $self->{_state_goal_p_func};
}
sub state_num_successors_func {
my $self = shift;
if (@_) { $self->{_state_num_successors_func} = shift }
return $self->{_state_num_successors_func};
}
sub state_successors_iterator {
my $self = shift;
if (@_) { $self->{_state_successors_iterator} = shift }
return $self->{_state_successors_iterator};
}
sub state_get_data_func {
my $self = shift;
if (@_) { $self->{_state_get_data_func} = shift }
return $self->{_state_get_data_func};
}
sub show_prog_func {
my $self = shift;
if (@_) { $self->{_show_prog_func} = shift }
return $self->{_show_prog_func};
}
###################################################################
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
# be multiple start-states.
#
###################################################################
sub add_start_state
{
my ($self, $state) = @_;
my $state_eval_func = $self->{_state_eval_func};
my $state_goal_p_func = $self->{_state_goal_p_func};
my $state_num_successors_func = $self->{_state_num_successors_func},
my $state_successors_iterator = $self->{_state_successors_iterator},
my $state_get_data_func = $self->{_state_get_data_func};
# make sure required functions have been defined
if(!defined($state_eval_func)){
croak "SMAstar: evaluation function is not defined\n";
}
if(!defined($state_goal_p_func)){
croak "SMAstar: goal function is not defined\n";
}
if(!defined($state_num_successors_func)){
croak "SMAstar: num successors function is not defined\n";
}
if(!defined($state_successors_iterator)){
croak "SMAstar: successor iterator is not defined\n";
}
# create a path object from this state
my $state_obj = AI::Pathfinding::SMAstar::Path->new(
_state => $state,
_eval_func => $state_eval_func,
_goal_p_func => $state_goal_p_func,
_num_successors_func => $state_num_successors_func,
_successors_iterator => $state_successors_iterator,
_get_data_func => $state_get_data_func,
);
my $fcost = AI::Pathfinding::SMAstar::Path::fcost($state_obj);
# check if the fcost of this node looks OK (is numeric)
unless(Scalar::Util::looks_like_number($fcost)){
croak "Error: f-cost of state is not numeric. Cannot add state to queue.\n";
}
$state_obj->f_cost($fcost);
# check if the num_successors function returns a number
my $num_successors = $state_obj->get_num_successors();
unless(Scalar::Util::looks_like_number($num_successors)){
croak "Error: Number of state successors is not numeric. Cannot add state to queue.\n";
}
# test out the iterator function to make sure it returns
# an object of the correct type
my $classname = ref($state);
my $test_successor_iterator = $state_obj->{_successors_iterator}->($state);
my $test_successor = $test_successor_iterator->($state);
my $succ_classname = ref($test_successor);
unless($succ_classname eq $classname){
croak "Error: Successor iterator method of object $classname does " .
"not return an object of type $classname.\n";
}
# add this node to the queue
$self->{_priority_queue}->insert($state_obj);
}
###################################################################
#
# start the SMAstar search process
#
###################################################################
sub start_search
{
my ($self,
$log_function,
$str_function,
$max_states_in_queue,
$max_cost,
) = @_;
if(!defined($str_function)){
croak "SMAstar start_search: str_function is not defined.\n";
}
sma_star_tree_search(\($self->{_priority_queue}),
\&AI::Pathfinding::SMAstar::Path::is_goal,
\&AI::Pathfinding::SMAstar::Path::get_descendants_iterator_smastar,
\&AI::Pathfinding::SMAstar::Path::fcost,
\&AI::Pathfinding::SMAstar::Path::backup_fvals,
$log_function,
$str_function,
\&AI::Pathfinding::SMAstar::Path::progress,
$self->{_show_prog_func},
$max_states_in_queue,
$max_cost,
);
}
#################################################################
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
#
#
#################################################################
sub sma_star_tree_search
{
my ($priority_queue,
$goal_p,
$successors_func,
$eval_func,
$backup_func,
$log_function, # debug string func; represent state object as a string.
$str_function,
$prog_function,
$show_prog_func,
$max_states_in_queue,
$max_cost,
) = @_;
my $iteration = 0;
my $num_states_in_queue = $$priority_queue->size();
my $max_extra_states_in_queue = $max_states_in_queue;
$max_states_in_queue = $num_states_in_queue + $max_extra_states_in_queue;
my $max_depth = ($max_states_in_queue - $num_states_in_queue);
my $best; # the best candidate for expansion
if($$priority_queue->is_empty() || !$$priority_queue){
return;
}
else{
my $num_successors = 0;
# loop over the elements in the priority queue
while(!$$priority_queue->is_empty()){
# determine the current size of the queue
my $num_states_in_queue = $$priority_queue->{_size};
# get the best candidate for expansion from the queue
$best = $$priority_queue->deepest_lowest_cost_leaf_dont_remove();
#------------------------------------------------------
if(!$DEBUG){
my $str = $log_function->($best);
$show_prog_func->($iteration, $num_states_in_queue, $str);
}
else{
my $str = $log_function->($best);
print "best is: " . $str_function->($best) . ", cost: " . $best->{_f_cost} . "\n";
}
#------------------------------------------------------
if($best->$goal_p()) {
# goal achieved! iteration: $iteration, number of
# states in queue: $num_states_in_queue.
return $best;
}
elsif($best->{_f_cost} >= $max_cost){
croak "\n\nSearch unsuccessful. max_cost reached (cost: $max_cost).\n";
}
else{
my $successors_iterator = $best->$successors_func();
my $succ = $successors_iterator->();
if($succ){
# if succ is at max depth and is not a goal node, set succ->fcost to infinity
if($succ->depth() >= $max_depth && !$succ->$goal_p() ){
$succ->{_f_cost} = $max_cost;
}
else{
# calling eval for comparison, and maintaining pathmax property
$succ->{_f_cost} = max($eval_func->($succ), $eval_func->($best));
my $descendant_index = $succ->{_descendant_index};
$best->{_descendant_fcosts}->[$descendant_index] = $succ->{_f_cost};
}
}
# determine if $best is completed, and if so backup values
if($best->is_completed()){
# remove from queue first, back up fvals, then insert back on queue.
# this way, it gets placed in its rightful place on the queue.
my $fval_before_backup = $best->{_f_cost};
# STEPS:
# 1) remove best and all antecedents from queue, but only if they are
# going to be altered by backing-up fvals. This is because
# removing and re-inserting in queue changes temporal ordering,
# and we don't want to do that unless the node will be
# placed in a new cost-bucket/tree.
# 2) then backup fvals
# 3) then re-insert best and all antecedents back on queue.
# Check if need for backup fvals
$best->check_need_fval_change();
my $cmp_func = sub {
my ($str) = @_;
return sub{
my ($obj) = @_;
my $obj_path_str = $str_function->($obj);
if($obj_path_str eq $str){
return 1;
}
else{
return 0;
}
}
};
my $antecedent = $best->{_antecedent};
my %was_on_queue;
my $i = 0;
# Now remove the offending nodes from queue, if any
if($best->need_fval_change()){
# remove best from the queue
$best = $$priority_queue->deepest_lowest_cost_leaf();
while($antecedent){
my $path_str = $str_function->($antecedent);
if($antecedent->is_on_queue() && $antecedent->need_fval_change()){
$was_on_queue{$i} = 1;
$$priority_queue->remove($antecedent, $cmp_func->($path_str));
}
$antecedent = $antecedent->{_antecedent};
$i++;
}
}
# Backup fvals
if($best->need_fval_change()){
$best->$backup_func();
}
# Put everything back on the queue
if($best->need_fval_change()){
$$priority_queue->insert($best);
my $antecedent = $best->{_antecedent};
my $i = 0;
while($antecedent){
if($was_on_queue{$i} && $antecedent->need_fval_change()){
# the antecedent needed fval change too.
$$priority_queue->insert($antecedent);
}
if($antecedent->need_fval_change()){
# set need_fval_change back to 0, so it will not be automatically seen as
# needing changed in the future. This is important, since we do not want
# to remove an element from the queue *unless* we need to change the fcost.
# This is because when we remove it from the queue and re-insert it, it
# loses its seniority in the queue (it becomes the newest node at its cost
# and depth) and will not be removed at the right time when searching for
# deepest_lowest_cost_leafs or shallowest_highest_cost_leafs.
$antecedent->{_need_fcost_change} = 0;
}
$antecedent = $antecedent->{_antecedent};
$i++;
}
# Again, set need_fval_change back to 0, so it will not be automatically
# seen as needing changed in the future.
$best->{_need_fcost_change} = 0;
}
}
#
# If best's descendants are all in memory, mark best as completed.
#
if($best->all_in_memory()) {
if(!($best->is_completed())){
$best->is_completed(1);
}
my $cmp_func = sub {
my ($str) = @_;
return sub{
my ($obj) = @_;
my $obj_str = $str_function->($obj);
if($obj_str eq $str){
return 1;
}
else{
return 0;
}
}
};
my $best_str = $str_function->($best);
# If best is not a root node
if($best->{_depth} != 0){
# descendant index is the unique index indicating which descendant
# this node is of its antecedent.
my $descendant_index = $best->{_descendant_index};
my $antecedent = $best->{_antecedent};
$$priority_queue->remove($best, $cmp_func->($best_str));
if($antecedent){
$antecedent->{_descendants_produced}->[$descendant_index] = 0;
}
}
}
# there are no more successors of $best
if(!$succ){
next;
}
my $antecedent;
my @antecedents_that_need_to_be_inserted;
# If the maximum number of states in the queue has been reached,
# we need to remove the shallowest-highest-cost leaf to make room
# for more nodes. That means we have to make sure that the antecedent
# produces this descendant again at some point in the future if needed.
if($num_states_in_queue > $max_states_in_queue){
my $shcl_obj = $$priority_queue->shallowest_highest_cost_leaf($best, $succ, $str_function);
if(!$shcl_obj){
croak "Error while pruning queue: shallowest-highest-cost-leaf was null\n";
}
$antecedent = $shcl_obj->{_antecedent};
if($antecedent){
my $antecedent_successors = \$antecedent->{_descendants_list};
$antecedent->remember_forgotten_nodes_fcost($shcl_obj);
$antecedent->{_forgotten_nodes_num} = $antecedent->{_forgotten_nodes_num} + 1;
my $descendant_index = $shcl_obj->{_descendant_index};
# record the index of this descendant in the forgotten_nodes list
$antecedent->{_forgotten_nodes_offsets}->{$descendant_index} = 1;
# flag the antecedent as not having this descendant in the queue
$antecedent->{_descendants_produced}->[$descendant_index] = 0;
$antecedent->{_descendant_fcosts}->[$descendant_index] = -1;
# flag the ancestor node as having deleted a descendant
$antecedent->descendants_deleted(1);
# update the number of descendants this node has in memory
$antecedent->{_num_successors_in_mem} = $antecedent->{_num_successors_in_mem} - 1;
# update the total number of nodes in the queue.
$num_states_in_queue--;
}
} # end if (num_states_on_queue > max_states)
# if there is a successor to $best, insert it in the priority queue.
if($succ){
$$priority_queue->insert($succ);
$best->{_num_successors_in_mem} = $best->{_num_successors_in_mem} + 1;
}
else{
croak "Error: no successor to insert\n";
}
}
}
continue {
$iteration++;
}
print "\n\nreturning unsuccessfully. iteration: $iteration\n";
return;
}
}
sub max
{
my ($n1, $n2) = @_;
return ($n1 > $n2 ? $n1 : $n2);
}
sub fp_compare {
my ($a, $b, $dp) = @_;
my $a_seq = sprintf("%.${dp}g", $a);
my $b_seq = sprintf("%.${dp}g", $b);
if($a_seq eq $b_seq){
return 0;
}
elsif($a_seq lt $b_seq){
return -1;
}
else{
return 1;
}
}
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
AI::Pathfinding::SMAstar - Simplified Memory-bounded A* Search
=head1 SYNOPSIS
use AI::Pathfinding::SMAstar;
=head2 EXAMPLE
##################################################################
#
# This example uses a hypothetical object called FrontierObj, and
# shows the functions that the FrontierObj class must feature in
# order to perform a path-search in a solution space populated by
# FrontierObj objects.
#
##################################################################
my $smastar = AI::Pathfinding::SMAstar->new(
# evaluates f(n) = g(n) + h(n), returns a number
_state_eval_func => \&FrontierObj::evaluate,
# when called on a node, returns 1 if it is a goal
_state_goal_p_func => \&FrontierObj::goal_test,
# must return the number of successors of a node
_state_num_successors_func => \&FrontierObj::get_num_successors,
# must return *one* successor at a time
_state_successors_iterator => \&FrontierObj::get_successors_iterator,
# can be any suitable string representation
_state_get_data_func => \&FrontierObj::string_representation,
# gets called once per iteration, useful for showing algorithm progress
_show_prog_func => \&FrontierObj::progress_callback,
);
# You can start the search from multiple start-states.
# Add the initial states to the smastar object before starting the search.
foreach my $frontierObj (@start_states){
$smastar->add_start_state($frontierObj);
}
#
# Start the search. If successful, $frontierGoalPath will contain the
# goal path. The optimal path to the goal node will be encoded in the
# ancestry of the goal path. $frontierGoalPath->antecedent() contains
# the goal path's parent path, and so forth back to the start path, which
# contains only the start state.
#
# $frontierGoalPath->state() contains the goal FrontierObj itself.
#
my $frontierGoalPath = $smastar->start_search(
\&log_function, # returns a string used for logging progress
\&str_function, # returns a string used to *uniquely* identify a node
$max_states_in_queue, # indicate the maximum states allowed in memory
$MAX_COST, # indicate the maximum cost allowed in search
);
In the example above, a hypothetical object, C<FrontierObj>, is used to
represent a state, or I<node> in your search space. To use SMA* search to
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
=head1 METHODS
=head2 new()
my $smastar = AI::Pathfinding::SMAstar->new();
Creates a new SMA* search object.
=head2 start_search()
my $frontierGoalObj = $smastar->start_search(
\&log_function, # returns a string used for logging progress
\&str_function, # returns a string used to *uniquely* identify a node
$max_states_in_queue, # indicate the maximum states allowed in memory
$MAX_COST, # indicate the maximum cost allowed in search
);
Initiates a memory-bounded search. When calling this function, pass a handle to
a function for recording current status( C<log_function> above- this can be
an empty subroutine if you don't care), a function that returns a *unique* string
representing a node in the search-space (this *cannot* be an empty subroutine), a
lib/AI/Pathfinding/SMAstar.pm view on Meta::CPAN
value (beyond which the search will cease).
=head2 state_eval_func()
$smastar->state_eval_func(\&FrontierObj::evaluate);
Set or get the handle to the function that returns the cost of the object
argument (node) in the search space.
=head2 state_goal_p_func()
$smastar->state_goal_p_func(\&FrontierObj::goal_test);
Set/get the handle to the goal predicate function. This is a function
that returns 1 if the argument object is a goal node, or 0 otherwise.
=head2 state_num_successors_func()
$smastar->state_num_successors_func(\&FrontierObj::get_num_successors);
Set/get the handle to the function that returns the number of successors
of this the object argument (node).
=head2 state_successors_iterator()
$smastar->state_successors_iterator(\&FrontierObj::get_successors_iterator);
Set/get the handle to the function that returns iterator that produces the
next successor of this node.
=head2 state_get_data_func()
$smastar->state_get_data_func(\&FrontierObj::string_representation);
Set/get the handle to the function that returns a string
representation of this node.
=head2 show_prog_func()
$smatar->show_prog_func(\&FrontierObj::progress_callback);
Sets/gets the callback function for displaying the progress of the search.
It can be an empty callback (sub{}) if you do not need this output.
=head2 DEPENDENCIES
Tree::AVL
Test::More
=head2 INCLUDED MODULES
AI::Pathfinding::SMAstar
AI::Pathfinding::SMAstar::Path
AI::Pathfinding::SMAstar::PriorityQueue
AI::Pathfinding::SMAstar::TreeOfQueues
=head2 EXPORT
AI-Perceptron-Simple
view release on metacpan or search on metacpan
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
use constant TUNE_UP => 1;
use constant TUNE_DOWN => 0;
=head1 SYNOPSIS
#!/usr/bin/perl
use AI::Perceptron::Simple qw(...);
# create a new nerve / neuron / perceptron
$nerve = AI::Perceptron::Simple->new( {
initial_value => $size_of_each_dendrite,
learning_rate => 0.3, # optional
threshold => 0.85, # optional
attribs => \@dendrites,
} );
# train
$nerve->tame( ... );
$nerve->exercise( ... );
$nerve->train( $training_data_csv, $expected_column_name, $save_nerve_to );
# or
$nerve->train(
$training_data_csv, $expected_column_name, $save_nerve_to,
$show_progress, $identifier); # these two parameters must go together
# validate
$nerve->take_lab_test( ... );
$nerve->take_mock_exam( ... );
# fill results to original file
$nerve->validate( {
stimuli_validate => $validation_data_csv,
predicted_column_index => 4,
} );
# or
# fill results to a new file
$nerve->validate( {
stimuli_validate => $validation_data_csv,
predicted_column_index => 4,
results_write_to => $new_csv
} );
# test - see "validate" method, same usage
$nerve->take_real_exam( ... );
$nerve->work_in_real_world( ... );
$nerve->test( ... );
# confusion matrix
my %c_matrix = $nerve->get_confusion_matrix( {
full_data_file => $file_csv,
actual_output_header => $header_name,
predicted_output_header => $predicted_header_name,
more_stats => 1, # optional
} );
# accessing the confusion matrix
my @keys = qw( true_positive true_negative false_positive false_negative
total_entries accuracy sensitivity );
for ( @keys ) {
print $_, " => ", $c_matrix{ $_ }, "\n";
}
# output to console
$nerve->display_confusion_matrix( \%c_matrix, {
zero_as => "bad apples", # cat milk green etc.
one_as => "good apples", # dog honey pink etc.
} );
# saving and loading data of perceptron locally
# NOTE: nerve data is automatically saved after each trainning process
use AI::Perceptron::Simple ":local_data";
my $nerve_file = "apples.nerve";
preserve( ... );
save_perceptron( $nerve, $nerve_file );
# load data of percpetron for use in actual program
my $apple_nerve = revive( ... );
my $apple_nerve = load_perceptron( $nerve_file );
# for portability of nerve data
use AI::Perceptron::Simple ":portable_data";
my $yaml_nerve_file = "pearls.yaml";
preserve_as_yaml ( ... );
save_perceptron_yaml ( $nerve, $yaml_nerve_file );
# load nerve data on the other computer
my $pearl_nerve = revive_from_yaml ( ... );
my $pearl_nerve = load_perceptron_yaml ( $yaml_nerve_file );
# processing data
use AI::Perceptron::Simple ":process_data";
shuffle_stimuli ( ... )
shuffle_data ( ORIGINAL_STIMULI, $new_file_1, $new_file_2, ... );
shuffle_data ( $original_stimuli => $new_file_1, $new_file_2, ... );
=head1 EXPORT
None by default.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
use Exporter qw( import );
our @EXPORT_OK = qw(
shuffle_data shuffle_stimuli
preserve save_perceptron revive load_perceptron
preserve_as_yaml save_perceptron_yaml revive_from_yaml load_perceptron_yaml
);
our %EXPORT_TAGS = (
process_data => [ qw( shuffle_data shuffle_stimuli ) ],
local_data => [ qw( preserve save_perceptron revive load_perceptron ) ],
portable_data => [ qw( preserve_as_yaml save_perceptron_yaml revive_from_yaml load_perceptron_yaml ) ],
);
=head1 DESCRIPTION
This module provides methods to build, train, validate and test a perceptron. It can also save the data of the perceptron for future use for any actual AI programs.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
Shuffles C<$original_data> or C<ORIGINAL_DATA> and saves them to other files.
=cut
sub shuffle_stimuli {
shuffle_data( @_ );
}
sub shuffle_data {
my $stimuli = shift or croak "Please specify the original file name";
my @shuffled_stimuli_names = @_
or croak "Please specify the output files for the shuffled data";
my @aoa;
for ( @shuffled_stimuli_names ) {
# copied from _real_validate_or_test
# open for shuffling
my $aoa = csv (in => $stimuli, encoding => ":encoding(utf-8)");
my $attrib_array_ref = shift @$aoa; # 'remove' the header, it's annoying :)
@aoa = shuffle( @$aoa ); # this can only process actual array
unshift @aoa, $attrib_array_ref; # put back the headers before saving file
csv( in => \@aoa, out => $_, encoding => ":encoding(utf-8)" )
and
print "Saved shuffled data into ", basename($_), "!\n";
}
}
=head1 CREATION RELATED SUBROUTINES/METHODS
=head2 new ( \%options )
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=back
=cut
sub new {
my $class = shift;
my $data_ref = shift;
my %data = %{ $data_ref };
# check keys
$data{ learning_rate } = LEARNING_RATE if not exists $data{ learning_rate };
$data{ threshold } = THRESHOLD if not exists $data{ threshold };
#####
# don't pack this key checking process into a subroutine for now
# this is also used in &_real_validate_or_test
my @missing_keys;
for ( qw( initial_value attribs ) ) {
push @missing_keys, $_ unless exists $data{ $_ };
}
croak "Missing keys: @missing_keys" if @missing_keys;
#####
# continue to process the rest of the data
my %attributes;
for ( @{ $data{ attribs } } ) {
$attributes{ $_ } = $data{ initial_value };
}
my %processed_data = (
learning_rate => $data{ learning_rate },
threshold => $data{ threshold },
attributes_hash_ref => \%attributes,
);
bless \%processed_data, $class;
}
=head2 get_attributes
Obtains a hash of all the attributes of the perceptron
=cut
sub get_attributes {
my $self = shift;
%{ $self->{attributes_hash_ref} };
}
=head2 learning_rate ( $value )
=head2 learning_rate
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
If C<$value> is given, sets the learning rate to C<$value>. If not, then it returns the learning rate.
=cut
sub learning_rate {
my $self = shift;
if ( @_ ) {
$self->{learning_rate} = shift;
} else {
$self->{learning_rate}
}
}
=head2 threshold ( $value )
=head2 threshold
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
If C<$value> is given, sets the threshold / passing rate to C<$value>. If not, then it returns the passing rate.
=cut
sub threshold {
my $self = shift;
if ( @_ ) {
$self->{ threshold } = shift;
} else {
$self->{ threshold };
}
}
=head1 TRAINING RELATED SUBROUTINES/METHODS
All the training methods here have the same parameters as the two actual C<train> method and they all do the same stuff. They are also used in the same way.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
If C<$display_stats> is specified ie. set to C<1>, then you B<MUST> specify the C<$identifier>. C<$identifier> is the column / header name that is used to identify a specific row of data in C<$stimuli_train_csv>.
=cut
sub tame {
train( @_ );
}
sub exercise {
train( @_ );
}
sub train {
my $self = shift;
my( $stimuli_train_csv, $expected_output_header, $save_nerve_to_file, $display_stats, $identifier ) = @_;
$display_stats = 0 if not defined $display_stats;
if ( $display_stats and not defined $identifier ) {
croak "Please specifiy a string for \$identifier if you are trying to display stats";
}
# CSV processing is all according to the documentation of Text::CSV
open my $data_fh, "<:encoding(UTF-8)", $stimuli_train_csv
or croak "Can't open $stimuli_train_csv: $!";
my $csv = Text::CSV->new( {auto_diag => 1, binary => 1} );
my $attrib = $csv->getline($data_fh);
$csv->column_names( $attrib );
# individual row
ROW: while ( my $row = $csv->getline_hr($data_fh) ) {
# print $row->{book_name}, " -> ";
# print $row->{$expected_output_header} ? "æ„æž—\n" : "é…丽优å“\n";
# calculate the output and fine tune parameters if necessary
while (1) {
my $output = _calculate_output( $self, $row );
#print "Sum = ", $output, "\n";
# $expected_output_header to be checked together over here
# if output >= threshold
# then category/result aka output is considered 1
# else output considered 0
# output expected/actual tuning
# 0 0 -
# 1 0 down
# 0 1 up
# 1 1 -
if ( ($output >= $self->threshold) and ( $row->{$expected_output_header} eq 0 ) ) {
_tune( $self, $row, TUNE_DOWN );
if ( $display_stats ) {
print $row->{$identifier}, "\n";
print " -> TUNED DOWN";
print " Old sum = ", $output;
print " Threshold = ", $self->threshold;
print " New Sum = ", _calculate_output( $self, $row ), "\n";
}
} elsif ( ($output < $self->threshold) and ( $row->{$expected_output_header} eq 1 ) ) {
_tune( $self, $row, TUNE_UP );
if ( $display_stats ) {
print $row->{$identifier}, "\n";
print " -> TUNED UP";
print " Old sum = ", $output;
print " Threshold = ", $self->threshold;
print " New Sum = ", _calculate_output( $self, $row ), "\n";
}
} elsif ( ($output < $self->threshold) and ( $row->{$expected_output_header} eq 0 ) ) {
if ( $display_stats ) {
print $row->{$identifier}, "\n";
print " -> NO TUNING NEEDED";
print " Sum = ", _calculate_output( $self, $row );
print " Threshold = ", $self->threshold, "\n";
}
next ROW;
} elsif ( ($output >= $self->threshold) and ( $row->{$expected_output_header} eq 1 ) ) {
if ( $display_stats ) {
print $row->{$identifier}, "\n";
print " -> NO TUNING NEEDED";
print " Sum = ", _calculate_output( $self, $row );
print " Threshold = ", $self->threshold, "\n";
}
next ROW;
} #else { print "Something's not right\n'" }
}
}
close $data_fh;
save_perceptron( $self, $save_nerve_to_file ); # this doesn't return anything
}
=head2 &_calculate_output( $self, \%stimuli_hash )
Calculates and returns the C<sum(weightage*input)> for each individual row of data. Actually, it justs add up all the existing weight since the C<input> is always 1 for now :)
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
This subroutine should be called in the procedural way for now.
=cut
sub _calculate_output {
my $self = shift;
my $stimuli_hash_ref = shift;
my %dendrites = $self->get_attributes;
my $sum; # this is the output
for ( keys %dendrites ) {
# if input is 1 for a dendrite, then calculate it
if ( $stimuli_hash_ref->{ $_ } ) {
# $sum += $dendrites{ $_ } * 1; # no need, if 1 then it is always the value itself
# this is very efficient, nice :)
$sum += $dendrites{ $_ };
}
}
$sum;
}
=head2 &_tune( $self, \%stimuli_hash, $tune_up_or_down )
Fine tunes the nerve. This will directly alter the attributes values in C<$self> according to the attributes / dendrites specified in C<new>.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
This subroutine should be called in the procedural way for now.
=cut
sub _tune {
my $self = shift;
my ( $stimuli_hash_ref, $tuning_status ) = @_;
my %dendrites = $self->get_attributes;
for ( keys %dendrites ) {
if ( $tuning_status == TUNE_DOWN ) {
if ( $stimuli_hash_ref->{ $_ } ) { # must check this one, it must be 1 before we can alter the actual dendrite size in the nerve :)
$self->{ attributes_hash_ref }{ $_ } -= $self->learning_rate;
}
#print $_, ": ", $self->{ attributes_hash_ref }{ $_ }, "\n";
} elsif ( $tuning_status == TUNE_UP ) {
if ( $stimuli_hash_ref->{ $_ } ) {
$self->{ attributes_hash_ref }{ $_ } += $self->learning_rate;
}
#print $_, ": ", $self->{ attributes_hash_ref }{ $_ }, "\n";
}
}
}
=head1 VALIDATION RELATED METHODS
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
I<*This method will call C<_real_validate_or_test> to do the actual work.>
=cut
sub take_mock_exam {
my ( $self, $data_hash_ref ) = @_;
$self->_real_validate_or_test( $data_hash_ref );
}
sub take_lab_test {
my ( $self, $data_hash_ref ) = @_;
$self->_real_validate_or_test( $data_hash_ref );
}
sub validate {
my ( $self, $data_hash_ref ) = @_;
$self->_real_validate_or_test( $data_hash_ref );
}
=head1 TESTING RELATED SUBROUTINES/METHODS
All the testing methods here have the same parameters as the actual C<test> method and they all do the same stuff. They are also used in the same way.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
# redirect to _real_validate_or_test
sub take_real_exam {
my ( $self, $data_hash_ref ) = @_;
$self->_real_validate_or_test( $data_hash_ref );
}
sub work_in_real_world {
my ( $self, $data_hash_ref ) = @_;
$self->_real_validate_or_test( $data_hash_ref );
}
sub test {
my ( $self, $data_hash_ref ) = @_;
$self->_real_validate_or_test( $data_hash_ref );
}
=head2 _real_validate_or_test ( $data_hash_ref )
This is where the actual validation or testing takes place.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
sub _real_validate_or_test {
my $self = shift; my $data_hash_ref = shift;
#####
my @missing_keys;
for ( qw( stimuli_validate predicted_column_index ) ) {
push @missing_keys, $_ unless exists $data_hash_ref->{ $_ };
}
croak "Missing keys: @missing_keys" if @missing_keys;
#####
my $stimuli_validate = $data_hash_ref->{ stimuli_validate };
my $predicted_index = $data_hash_ref->{ predicted_column_index };
# actual processing starts here
my $output_file = defined $data_hash_ref->{ results_write_to }
? $data_hash_ref->{ results_write_to }
: $stimuli_validate;
# open for writing results
my $aoa = csv (in => $stimuli_validate, encoding => ":encoding(utf-8)");
my $attrib_array_ref = shift @$aoa; # 'remove' the header, it's annoying :)
$aoa = _fill_predicted_values( $self, $stimuli_validate, $predicted_index, $aoa );
# put back the array of headers before saving file
unshift @$aoa, $attrib_array_ref;
print "Saving data to $output_file\n";
csv( in => $aoa, out => $output_file, encoding => ":encoding(utf-8)" );
print "Done saving!\n";
}
=head2 &_fill_predicted_values ( $self, $stimuli_validate, $predicted_index, $aoa )
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
This subroutine should be called in the procedural way.
=cut
sub _fill_predicted_values {
my ( $self, $stimuli_validate, $predicted_index, $aoa ) = @_;
# CSV processing is all according to the documentation of Text::CSV
open my $data_fh, "<:encoding(UTF-8)", $stimuli_validate
or croak "Can't open $stimuli_validate: $!";
my $csv = Text::CSV->new( {auto_diag => 1, binary => 1} );
my $attrib = $csv->getline($data_fh);
$csv->column_names( $attrib );
# individual row
my $row = 0;
while ( my $data = $csv->getline_hr($data_fh) ) {
if ( _calculate_output( $self, $data ) >= $self->threshold ) {
# write 1 into aoa
$aoa->[ $row ][ $predicted_index ] = 1;
} else {
#write 0 into aoa
$aoa->[ $row ][ $predicted_index ] = 0;
}
$row++;
}
close $data_fh;
$aoa;
}
=head1 RESULTS RELATED SUBROUTINES/METHODS
This part is related to generating the confusion matrix.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
sub get_exam_results {
my ( $self, $info ) = @_;
$self->get_confusion_matrix( $info );
}
sub get_confusion_matrix {
my ( $self, $info ) = @_;
my %c_matrix = _collect_stats( $info ); # processes total_entries, accuracy, sensitivity etc
%c_matrix;
}
=head2 &_collect_stats ( \%options )
Generates a hash of confusion matrix based on C<%options> given in the C<get_confusion_matrix> method.
=cut
sub _collect_stats {
my $info = shift;
my $file = $info->{ full_data_file };
my $actual_header = $info->{ actual_output_header };
my $predicted_header = $info->{ predicted_output_header };
my $more_stats = defined ( $info->{ more_stats } ) ? 1 : 0;
my %c_matrix = (
true_positive => 0, true_negative => 0, false_positive => 0, false_negative => 0,
accuracy => 0, sensitivity => 0
);
# CSV processing is all according to the documentation of Text::CSV
open my $data_fh, "<:encoding(UTF-8)", $file
or croak "Can't open $file: $!";
my $csv = Text::CSV->new( {auto_diag => 1, binary => 1} );
my $attrib = $csv->getline($data_fh); # get the row of headers, can't specify any column
# shouldn't be a problem, since we're reading line by line :)
$csv->column_names( $attrib );
# individual row
while ( my $row = $csv->getline_hr($data_fh) ) {
# don't pack this part into another subroutine, number of rows can be very big
if ( $row->{ $actual_header } == 1 and $row->{ $predicted_header } == 1 ) {
# true positive
$c_matrix{ true_positive }++;
} elsif ( $row->{ $actual_header } == 0 and $row->{ $predicted_header } == 0 ) {
# true negative
$c_matrix{ true_negative }++;
} elsif ( $row->{ $actual_header } == 1 and $row->{ $predicted_header } == 0 ) {
# false negative
$c_matrix{ false_negative }++;
} elsif ( $row->{ $actual_header } == 0 and $row->{ $predicted_header } == 1 ) {
# false positive
$c_matrix{ false_positive }++;
} else {
croak "Something's wrong!\n".
"Make sure that the actual and predicted values in your file are binary ie 0 or 1" ;
}
}
close $data_fh;
_calculate_total_entries( \%c_matrix );
_calculate_sensitivity( \%c_matrix );
_calculate_accuracy( \%c_matrix );
if ( $more_stats == 1 ) {
_calculate_precision( \%c_matrix );
_calculate_specificity( \%c_matrix );
_calculate_f1_score( \%c_matrix );
# unimplemented, some more left
_calculate_negative_predicted_value( \%c_matrix ); #
_calculate_false_negative_rate( \%c_matrix ); #
_calculate_false_positive_rate( \%c_matrix ); #
_calculate_false_discovery_rate( \%c_matrix ); #
_calculate_false_omission_rate( \%c_matrix ); #
_calculate_balanced_accuracy( \%c_matrix ); #
}
%c_matrix;
}
=head2 &_calculate_total_entries ( $c_matrix_ref )
Calculates and adds the data for the C<total_entries> key in the confusion matrix hash.
=cut
sub _calculate_total_entries {
my $c_matrix = shift;
my $total = $c_matrix->{ true_negative } + $c_matrix->{ false_positive };
$total += $c_matrix->{ false_negative } + $c_matrix->{ true_positive };
$c_matrix->{ total_entries } = $total;
}
=head2 &_calculate_accuracy ( $c_matrix_ref )
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
sub _calculate_accuracy {
my $c_matrix = shift;
my $numerator = $c_matrix->{ true_positive } + $c_matrix->{ true_negative };
my $denominator = $numerator + $c_matrix->{ false_positive } + $c_matrix->{ false_negative };
$c_matrix->{ accuracy } = $numerator / $denominator * 100;
# no need to return anything, we're using ref
}
=head2 &_calculate_sensitivity ( $c_matrix_ref )
Calculates and adds the data for the C<sensitivity> key in the confusion matrix hash.
=cut
sub _calculate_sensitivity {
my $c_matrix = shift;
my $numerator = $c_matrix->{ true_positive };
my $denominator = $numerator + $c_matrix->{ false_negative };
$c_matrix->{ sensitivity } = $numerator / $denominator * 100;
# no need to return anything, we're using ref
}
=head2 &_calculate_precision ( $c_matrix_ref )
Calculates and adds the data for the C<precision> key in the confusion matrix hash.
=cut
sub _calculate_precision {
my $c_matrix = shift;
my $numerator = $c_matrix->{ true_positive };
my $denominator = $numerator + $c_matrix->{ false_positive };
$c_matrix->{ precision } = $numerator / $denominator * 100;
}
=head2 &_calculate_specificity ( $c_matrix_ref )
Calculates and adds the data for the C<specificity> key in the confusion matrix hash.
=cut
sub _calculate_specificity {
my $c_matrix = shift;
my $numerator = $c_matrix->{ true_negative };
my $denominator = $numerator + $c_matrix->{ false_positive };
$c_matrix->{ specificity } = $numerator / $denominator * 100;
}
=head2 &_calculate_f1_score ( $c_matrix_ref )
Calculates and adds the data for the C<F1_Score> key in the confusion matrix hash.
=cut
sub _calculate_f1_score {
my $c_matrix = shift;
my $numerator = 2 * $c_matrix->{ true_positive };
my $denominator = $numerator + $c_matrix->{ false_positive } + $c_matrix->{ false_negative };
$c_matrix->{ F1_Score } = $numerator / $denominator * 100;
}
=head2 &_calculate_negative_predicted_value( $c_matrix_ref )
Calculates and adds the data for the C<negative_predicted_value> key in the confusion matrix hash.
=cut
sub _calculate_negative_predicted_value {
my $c_matrix = shift;
my $numerator = $c_matrix->{ true_negative };
my $denominator = $numerator + $c_matrix->{ false_negative };
$c_matrix->{ negative_predicted_value } = $numerator / $denominator * 100;
}
=head2 &_calculate_false_negative_rate( $c_matrix_ref )
Calculates and adds the data for the C<false_negative_rate> key in the confusion matrix hash.
=cut
sub _calculate_false_negative_rate {
my $c_matrix = shift;
my $numerator = $c_matrix->{ false_negative };
my $denominator = $numerator + $c_matrix->{ true_positive };
$c_matrix->{ false_negative_rate } = $numerator / $denominator * 100;
}
=head2 &_calculate_false_positive_rate( $c_matrix_ref )
Calculates and adds the data for the C<false_positive_rate> key in the confusion matrix hash.
=cut
sub _calculate_false_positive_rate {
my $c_matrix = shift;
my $numerator = $c_matrix->{ false_positive };
my $denominator = $numerator + $c_matrix->{ true_negative };
$c_matrix->{ false_positive_rate } = $numerator / $denominator * 100;
}
=head2 &_calculate_false_discovery_rate( $c_matrix_ref )
Calculates and adds the data for the C<false_discovery_rate> key in the confusion matrix hash.
=cut
sub _calculate_false_discovery_rate {
my $c_matrix = shift;
my $numerator = $c_matrix->{ false_positive };
my $denominator = $numerator + $c_matrix->{ true_positive };
$c_matrix->{ false_discovery_rate } = $numerator / $denominator * 100;
}
=head2 &_calculate_false_omission_rate( $c_matrix_ref )
Calculates and adds the data for the C<false_omission_rate> key in the confusion matrix hash.
=cut
sub _calculate_false_omission_rate {
my $c_matrix = shift;
my $numerator = $c_matrix->{ false_negative };
my $denominator = $numerator + $c_matrix->{ true_negative };
$c_matrix->{ false_omission_rate } = $numerator / $denominator * 100;
}
=head2 &_calculate_balanced_accuracy( $c_matrix_ref )
Calculates and adds the data for the C<balanced_accuracy> key in the confusion matrix hash.
=cut
sub _calculate_balanced_accuracy {
my $c_matrix = shift;
my $numerator = $c_matrix->{ sensitivity } + $c_matrix->{ specificity };
my $denominator = 2;
$c_matrix->{ balanced_accuracy } = $numerator / $denominator; # numerator already in %
}
=head2 display_exam_results ( ... )
The parameters are the same as C<display_confusion_matrix>. See the next method.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
sub display_exam_results {
my ( $self, $c_matrix, $labels ) = @_;
$self->display_confusion_matrix( $c_matrix, $labels );
}
sub display_confusion_matrix {
my ( $self, $c_matrix, $labels ) = @_;
#####
my @missing_keys;
for ( qw( zero_as one_as ) ) {
push @missing_keys, $_ unless exists $labels->{ $_ };
}
croak "Missing keys: @missing_keys" if @missing_keys;
#####
_print_extended_matrix ( _build_matrix( $c_matrix, $labels ) );
}
=head2 &_build_matrix ( $c_matrix, $labels )
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
sub _build_matrix {
my ( $c_matrix, $labels ) = @_;
my $predicted_columns = [ "P: ".$labels->{ zero_as }, "P: ".$labels->{ one_as }, "Sum" ];
my $actual_rows = [ "A: ".$labels->{ zero_as }, "A: ".$labels->{ one_as }, "Sum"];
# row sum
my $actual_0_sum = $c_matrix->{ true_negative } + $c_matrix->{ false_positive };
my $actual_1_sum = $c_matrix->{ false_negative } + $c_matrix->{ true_positive };
# column sum
my $predicted_0_sum = $c_matrix->{ true_negative } + $c_matrix->{ false_negative };
my $predicted_1_sum = $c_matrix->{ false_positive } + $c_matrix->{ true_positive };
my $data = [
[ $c_matrix->{ true_negative }, $c_matrix->{ false_positive }, $actual_0_sum ],
[ $c_matrix->{ false_negative }, $c_matrix->{ true_positive }, $actual_1_sum ],
[ $predicted_0_sum, $predicted_1_sum, $c_matrix->{ total_entries } ],
];
my $matrix = Text::Matrix->new(
rows => $actual_rows,
columns => $predicted_columns,
data => $data,
);
$matrix, $c_matrix;
}
=head2 &_print_extended_matrix ( $matrix, $c_matrix )
Extends and outputs the matrix on the screen.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=cut
sub _print_extended_matrix {
my ( $matrix, $c_matrix ) = @_;
print "~~" x24, "\n";
print "CONFUSION MATRIX (A:actual P:predicted)\n";
print "~~" x24, "\n";
print $matrix->matrix();
print "~~" x24, "\n";
print "Total of ", $c_matrix->{ total_entries } , " entries\n";
print " Accuracy: $c_matrix->{ accuracy } %\n";
print " Sensitivity: $c_matrix->{ sensitivity } %\n";
# more stats
print " Precision: $c_matrix->{ precision } %\n" if exists $c_matrix->{ precision };
print " Specificity: $c_matrix->{ specificity } %\n" if exists $c_matrix->{ specificity };
print " F1 Score: $c_matrix->{ F1_Score } %\n" if exists $c_matrix->{ F1_Score };
print " Negative Predicted Value: $c_matrix->{ negative_predicted_value } %\n" if exists $c_matrix->{ negative_predicted_value };
print " False Negative Rate: $c_matrix->{ false_negative_rate } %\n" if exists $c_matrix->{ false_negative_rate };
print " False Positive Rate: $c_matrix->{ false_positive_rate } %\n" if exists $c_matrix->{ false_positive_rate };
print " False Discovery Rate: $c_matrix->{ false_discovery_rate } %\n" if exists $c_matrix->{ false_discovery_rate };
print " False Omission Rate: $c_matrix->{ false_omission_rate } %\n" if exists $c_matrix->{ false_omission_rate };
print " Balanced Accuracy: $c_matrix->{ balanced_accuracy } %\n" if exists $c_matrix->{ balanced_accuracy };
print "~~" x24, "\n";
}
=head1 NERVE DATA RELATED SUBROUTINES
This part is about saving the data of the nerve. These subroutines can be imported using the C<:local_data> tag.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
process this will be called automatically.
=cut
sub preserve {
save_perceptron( @_ );
}
sub save_perceptron {
my $self = shift;
my $nerve_file = shift;
use Storable;
store $self, $nerve_file;
no Storable;
}
=head2 revive (...)
The parameters and usage are the same as C<load_perceptron>. See the next subroutine.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
Loads the data and turns it into a C<AI::Perceptron::Simple> object as the return value.
=cut
sub revive {
load_perceptron( @_ );
}
sub load_perceptron {
my $nerve_file_to_load = shift;
use Storable;
my $loaded_nerve = retrieve( $nerve_file_to_load );
no Storable;
$loaded_nerve;
}
=head1 NERVE PORTABILITY RELATED SUBROUTINES
These subroutines can be imported using the C<:portable_data> tag.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
Saves the C<AI::Perceptron::Simple> object into a C<YAML> file.
=cut
sub preserve_as_yaml {
save_perceptron_yaml( @_ );
}
sub save_perceptron_yaml {
my $self = shift;
my $nerve_file = shift;
use YAML;
YAML::DumpFile( $nerve_file, $self );
no YAML;
}
=head2 revive_from_yaml (...)
The parameters and usage are the same as C<load_perceptron>. See the next subroutine.
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
Loads the YAML data and turns it into a C<AI::Perceptron::Simple> object as the return value.
=cut
sub revive_from_yaml {
load_perceptron_yaml( @_ );
}
sub load_perceptron_yaml {
my $nerve_file_to_load = shift;
use YAML;
local $YAML::LoadBlessed = 1;
my $loaded_nerve = YAML::LoadFile( $nerve_file_to_load );
no YAML;
$loaded_nerve;
}
=head1 TO DO
These are the to-do's that B<MIGHT> be done in the future. Don't put too much hope in them please :)
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
=head1 SUPPORT
You can find documentation for this module with the perldoc command.
perldoc AI::Perceptron::Simple
You can also look for information at:
=over 4
lib/AI/Perceptron/Simple.pm view on Meta::CPAN
This software is Copyright (c) 2021 by Raphael Jong Jun Jie.
This is free software, licensed under:
The Artistic License 2.0 (GPL Compatible)
=cut
1; # End of AI::Perceptron::Simple
AI-Perceptron
view release on metacpan or search on metacpan
examples/and.pl view on Meta::CPAN
use Data::Dumper;
use AI::Perceptron;
print( "Example: training a perceptron to recognize an 'AND' function.\n",
"usage: $0 [<threshold> <weight1> <weight2>]\n" );
my $p = AI::Perceptron->new
->num_inputs( 2 )
->learning_rate( 0.1 );
if (@ARGV) {
$p->threshold( shift(@ARGV) )
->weights([ shift(@ARGV), shift(@ARGV) ]);
}
my @training_exs = (
[-1 => -1, -1],
[-1 => 1, -1],
[-1 => -1, 1],
[ 1 => 1, 1],
);
print "\nBefore Training\n";
dump_perceptron( $p );
print "\nTraining...\n";
examples/and.pl view on Meta::CPAN
print "\nAfter Training\n";
dump_perceptron( $p );
sub dump_perceptron {
my $p = shift;
print "\tThreshold: ", $p->threshold, " Weights: ", join(', ', @{ $p->weights }), "\n";
foreach my $inputs (@training_exs) {
my $target = $inputs->[0];
print "\tInputs = {", join(',', @$inputs[1..2]), "}, target=$target, output=", $p->compute_output( @$inputs[1..2] ), "\n";
}
}
AI-PredictionClient-Alien-TensorFlowServingProtos
view release on metacpan or search on metacpan
lib/AI/PredictionClient/Alien/TensorFlowServingProtos.pm view on Meta::CPAN
=head1 SYNOPSIS
In your Build.PL:
use Module::Build;
use AI::PredictionClient::Alien::TensorFlowServingProtos;
my $builder = Module::Build->new(
...
configure_requires => {
'AI::PredictionClient::Alien::TensorFlowServingProtos' => '0',
...
},
extra_compiler_flags => AI::PredictionClient::Alien::TensorFlowServingProtos->cflags,
extra_linker_flags => AI::PredictionClient::Alien::TensorFlowServingProtos->libs,
...
);
$build->create_build_script;
In your Makefile.PL:
use ExtUtils::MakeMaker;
use Config;
use AI::PredictionClient::Alien::TensorFlowServingProtos;
WriteMakefile(
...
CONFIGURE_REQUIRES => {
'AI::PredictionClient::Alien::TensorFlowServingProtos' => '0',
},
CCFLAGS => AI::PredictionClient::Alien::TensorFlowServingProtos->cflags . " $Config{ccflags}",
LIBS => [ AI::PredictionClient::Alien::TensorFlowServingProtos->libs ],
...
);
=cut
=head1 DESCRIPTION
lib/AI/PredictionClient/Alien/TensorFlowServingProtos.pm view on Meta::CPAN
module will download and build a private copy.
The system dependencies needed for this module to build are most often already installed.
If not, the following dependencies need to be installed.
$ [sudo] apt-get install build-essential make g++
See the Alien::Google::GRPC for potential additional build dependencies.
At this time only Linux builds are supported.
AI-PredictionClient
view release on metacpan or search on metacpan
bin/Inception.pl view on Meta::CPAN
my $default_port = '9000';
my $default_model = 'inception';
my $default_model_signature = 'predict_images';
option image_file => (
is => 'ro',
required => 1,
format => 's',
doc => '* Required: Path to image to be processed'
);
option host => (
is => 'ro',
required => 0,
format => 's',
default => $default_host,
doc => "IP address of the server [Default: $default_host]"
);
option port => (
is => 'ro',
required => 0,
format => 's',
default => $default_port,
doc => "Port number of the server [Default: $default_port]"
);
option model_name => (
is => 'ro',
required => 0,
format => 's',
default => $default_model,
doc => "Model to process image [Default: $default_model]"
);
option model_signature => (
is => 'ro',
required => 0,
format => 's',
default => $default_model_signature,
doc => "API signature for model [Default: $default_model_signature]"
);
option debug_verbose => (is => 'ro', doc => 'Verbose output');
option debug_loopback_interface => (
is => 'ro',
required => 0,
doc => "Test loopback through dummy server"
);
option debug_camel => (
is => 'ro',
required => 0,
doc => "Test using camel image"
);
sub run {
my ($self) = @_;
my $image_ref = $self->read_image($self->image_file);
my $client = AI::PredictionClient::InceptionClient->new(
host => $self->host,
port => $self->port
);
$client->model_name($self->model_name);
$client->model_signature($self->model_signature);
$client->debug_verbose($self->debug_verbose);
$client->loopback($self->debug_loopback_interface);
$client->camel($self->debug_camel);
printf("Sending image %s to server at host:%s port:%s\n",
$self->image_file, $self->host, $self->port);
if ($client->call_inception($image_ref)) {
my $results_ref = $client->inception_results;
my $classifications_ref = $results_ref->{'classes'};
my $scores_ref = $results_ref->{'scores'};
my $comments = 'Clasification Results for ' . $self->image_file;
my $results_text
= form
'.===========================================================================.',
'| Class | Score |',
'|-----------------------------------------------------------+---------------|',
'| {[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[} |{]].[[[[[[[[} |',
$classifications_ref, $scores_ref,
'|===========================================================================|',
'| {[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[} |',
$comments,
"'==========================================================================='";
print $results_text;
} else {
printf("Failed. Status: %s, Status Code: %s, Status Message: %s \n",
$client->status, $client->status_code, $client->status_message);
return 1;
}
return 0;
}
sub read_image {
my $self = shift;
return \'' if $self->debug_camel;
my $file_name = shift;
my $max_file_size = 16 * 1000 * 1000; # A large but safe maximum
open(my $fh, '<:raw', $file_name)
or die "Could not open file: $file_name";
read($fh, my $buffer, $max_file_size);
close $fh;
return \$buffer;
}
exit main->new_with_options->run;
__END__
AI-Prolog
view release on metacpan or search on metacpan
examples/append.pl view on Meta::CPAN
END_PROLOG
print "Appending two lists 'append([a],[b,c,d],Z).'\n";
$prolog->query('append([a],[b,c,d],Z).');
while (my $result = $prolog->results) {
print Dumper($result),"\n";
}
print "\nWhich lists appends to a known list to form another known list?\n'append(X,[b,c,d],[a,b,c,d]).'\n";
$prolog->query('append(X,[b,c,d],[a,b,c,d]).');
while (my $result = $prolog->results) {
print Dumper($result),"\n";
}
print "\nWhich lists can be appended to form a given list?\n'append(X, Y, [foo, bar, 7, baz]).'\n";
my $list = $prolog->list(qw/foo bar 7 baz/);
$prolog->query("append(X,Y,[$list]).");
while (my $result = $prolog->results) {
print Dumper($result),"\n";
}
( run in 0.899 second using v1.01-cache-2.11-cpan-39bf76dae61 )