AI::FANN(3pm)

📦

version 0.10 · 6 versions

💖 NAME 💖

AI::FANN (that's this page — AI::FANN(3pm) bestie 💅) is the Perl wrapper for the Fast Artificial Neural Network library. Neural nets, in Perl, slay.

✨ SYNOPSIS ✨

🧠 Train... (the glow-up era)

  use AI::FANN qw(:all);

  # create an ANN with 2 inputs, a hidden layer with 3 neurons and an
  # output layer with 1 neuron:
  my $ann = AI::FANN->new_standard(2, 3, 1);

  $ann->hidden_activation_function(FANN_SIGMOID_SYMMETRIC);
  $ann->output_activation_function(FANN_SIGMOID_SYMMETRIC);

  # create the training data for a XOR operator:
  my $xor_train = AI::FANN::TrainData->new( [-1, -1], [-1],
                                            [-1, 1], [1],
                                            [1, -1], [1],
                                            [1, 1], [-1] );

  $ann->train_on_data($xor_train, 500000, 1000, 0.001);

  $ann->save("xor.ann");

🔮 Run... (the main character era)

  use AI::FANN;

  my $ann = AI::FANN->new_from_file("xor.ann");

  for my $a (-1, 1) {
    for my $b (-1, 1) {
      my $out = $ann->run([$a, $b]);
      printf "xor(%f, %f) = %f\n", $a, $b, $out->[0];
    }
  }

🌸 DESCRIPTION 🌸

🚨 WARNING: THIS IS A VERY EARLY RELEASE, MAY CONTAIN CRITICAL BUGS!!! 🚨

okay so, deep breath — AI::FANN is a Perl wrapper for the Fast Artificial Neural Network (FANN) Library, which lives at http://fann.sourceforge.net. Here's the library describing itself, no cap:

Fast Artificial Neural Network Library is a free open source neural network library, which implements multilayer artificial neural networks in C with support for both fully connected and sparsely connected networks. Cross-platform execution in both fixed and floating point are supported. It includes a framework for easy handling of training data sets. It is easy to use, versatile, well documented, and fast. PHP, C++, .NET, Python, Delphi, Octave, Ruby, Pure Data and Mathematica bindings are available. A reference manual accompanies the library with examples and recommendations on how to use the library. A graphical user interface is also available for the library.

Literally iconic. 😤

The AI::FANN object oriented interface is an almost one-to-one map onto the C library API — but a few things got a lil makeover to be more perlish, and honestly? It ate. 💅

Two classes are in the group chat: AI::FANN, which wraps the C struct fann type, and AI::FANN::TrainData, which wraps struct fann_train_data.
Prefixes and the common parts of C function names that point at those structs got yeeted. So for example C's fann_train_data_shuffle becomes AI::FANN::TrainData::shuffle, which you'll usually call like...

  $train_data->shuffle;

Pairs of C get/set functions get wrapped in Perl as dual accessor methods named after the attribute, with no set_/get_ prefix. So:

  $ann->bit_fail_limit($limit); # sets the bit_fail_limit

  $bfl = $ann->bit_fail_limit;  # gets the bit_fail_limit

And pairs that also need extra indexing arguments? Also wrapped in dual accessors:

  # sets:
  $ann->neuron_activation_function($layer_ix, $neuron_ix, $actfunc);

  # gets:
  $af = $ann->neuron_activation_function($layer_ix, $neuron_ix);

⚠️ Important: on the Perl side, that optional value argument gets moved to the last position — on the C version of the set_ method it's usually the second argument. Don't get caught slippin'. 🙈

Some functions got renamed so the naming is more consistent and follows Perl conventions:
- fann_create_from_file => new_from_file
- fann_create_standard => new_standard
- fann_get_num_input => num_inputs
- fann_get_activation_function => neuron_activation_function
- fann_set_activation_function => ^^^ (yep, same as the one above — also neuron_activation_function)
- fann_set_activation_function_layer => layer_activation_function
- fann_set_activation_function_hidden => hidden_activation_function
- fann_set_activation_function_output => output_activation_function
Boolean methods return true on success and undef on failure.
Any error the C side reports gets automatically converted into a Perl exception. ✨ No manual error checking required after calling FANN functions — she handles it.
Memory management is automatic. You do not need to call destroy methods. Self-care. 🧖
Doubles are used for computations. Using floats or fixed point types is not supported. Periodt.

🎀 CONSTANTS 🎀

All the constants defined in the C documentation are exported from the module:

  # import all...
  use AI::FANN ':all';

  # or individual constants...
  use AI::FANN qw(FANN_TRAIN_INCREMENTAL FANN_GAUSSIAN);

Little detail we love: the values returned from these constant subs yield the integer value in numerical context, and the constant name when used as strings. Best of both worlds, fr fr. 🫶

The constants available are:

  # enum fann_train_enum:
  FANN_TRAIN_INCREMENTAL
  FANN_TRAIN_BATCH
  FANN_TRAIN_RPROP
  FANN_TRAIN_QUICKPROP

  # enum fann_activationfunc_enum:
  FANN_LINEAR
  FANN_THRESHOLD
  FANN_THRESHOLD_SYMMETRIC
  FANN_SIGMOID
  FANN_SIGMOID_STEPWISE
  FANN_SIGMOID_SYMMETRIC
  FANN_SIGMOID_SYMMETRIC_STEPWISE
  FANN_GAUSSIAN
  FANN_GAUSSIAN_SYMMETRIC
  FANN_GAUSSIAN_STEPWISE
  FANN_ELLIOT
  FANN_ELLIOT_SYMMETRIC
  FANN_LINEAR_PIECE
  FANN_LINEAR_PIECE_SYMMETRIC
  FANN_SIN_SYMMETRIC
  FANN_COS_SYMMETRIC
  FANN_SIN
  FANN_COS

  # enum fann_errorfunc_enum:
  FANN_ERRORFUNC_LINEAR
  FANN_ERRORFUNC_TANH

  # enum fann_stopfunc_enum:
  FANN_STOPFUNC_MSE
  FANN_STOPFUNC_BIT

...and there's more coming, bestie, don't go anywhere — this page is long and it's literally still going. 🌈

$ann->num_neurons — returns a list with the number of neurons on every layer 👑

AI::FANN::TrainData 🧚

This one wraps the C struct fann_train_data, and it's got exactly one job but it does it cute:

AI::FANN::TrainData->new_from_file($filename) — the <italic>-</italic> special, bestie: it's a straight-up wrapper around the C call, so the C documentation is where the full tea lives 📖
AI::FANN::TrainData->new($input1, $output1 [, $input2, $output2, ...]) — here $inputx and $outputx are arrays holding the values of the input and output layers, so you pass them in pairs, like a group chat where nobody gets left on read 💅
AI::FANN::TrainData->new_empty($num_data, $num_inputs, $num_outputs) — hands you a brand-new AI::FANN::TrainData object sized exactly like the args say. Heads up though: the initial values inside that object are random, and you 100% need to set them before you use this train data object to actually train an ANN. Not optional. Do not skip 🙈

Methods 🌸

$train->data($index) — returns two arrays: the input-layer values and the output-layer values, respectively, for that index 🫶
$train->data($index, $input, $output) — here $input and $output are two arrays, and the input and output layers at $index get set to the values in those arrays ✨
$train->shuffle — the <italic>-</italic> treatment again; C docs have the full description, bestie
$train->scale_input($new_min, $new_max) — same <italic>-</italic> situation, go check the C documentation
$train->scale_output($new_min, $new_max) — <italic>-</italic>, C docs, you know the drill 💖
$train->scale($new_min, $new_max) — <italic>-</italic>, straight to the C docs
$train->subset($pos, $length) — <italic>-</italic>, the C documentation is the source of truth here
$train->num_inputs — <italic>-</italic>, wrapped C call, C docs, mwah 💋
$train->num_outputs — <italic>-</italic>, again with the C docs
$train->length — <italic>-</italic>, and once more: C documentation 🔥

🛠️ INSTALLATION 🛠️

Go read the README file for the instructions on installing this module, bestie. That's where the whole how-to lives 📚

🐛 BUGS 🐛

Okay so, the honest tea:

Only tested on Linux 🐧
I/O is not performed through PerlIO, because the C library just doesn't have the infrastructure required to do that. It's giving "we tried, the C library said no" 💀
Send bug reports to my email address, or use the CPAN RT system 🫶

fr fr, if you find a bug, tell someone — don't just sit on it like an unread message 🙈

👀 SEE ALSO 👀

FANN homepage at <http://leenissen.dk/fann/index.php> — go look, it's iconic 🦋

⚖️ COPYRIGHT AND LICENSE ⚖️

This Perl module is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.8 or, at your option, any later version of Perl 5 you may have available. Slay 💅

Distributed under the GNU Lesser General Public License. 🎀

And that's the whole page, bestie!! This was perl v5.42.3, dated 2026-08-19, from the AI::FANN(3pm) manual — drink some water, go train a neural net, and remember: the C documentation is always there for you when the <italic>-</italic> leaves you hanging 💖✨🎀🦋

📦 where this page comes from

ubuntu stonking: libai-fann-perl 0.10-5build8 · Artistic-1.0-Perl OR GPL-1.0-or-later
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-11-03: a language model's rewrite of the original, under its terms