Fast Artificial Neural Network Library is a software library that enables the creation of multi-layer artificial neural networks.
This software tool includes a framework that makes it incredibly easy to handle training data sets. It features easy-to-use functionality with the ability to create, train and run an ANN with just three function calls while remaining fast and efficient.
In addition, it offers several features, including backpropagation training (RPROP, Quickprop, Batch, Incremental), evolving topology training which dynamically builds and trains the ANN (Cascade2), and several different activation functions.
One of its most notable features is its fast execution, which offers up to 150 times faster execution than other libraries. It's also incredibly flexible as it allows users to adjust many parameters and features on-the-fly.
The software comes with an easy-to-use reference manual, a 50+ page university report describing implementation considerations, and an introduction article, all of which provide density documentation. It's cross-platform compatible, with a configure script for linux and unix, dll files for windows, project files for MSVC++ and Borland compilers.
Ease of use is further enhanced by the availability of several easy-to-use examples, such as simple train and simple test examples. The software also makes it easy to save and load entire ANNs while enabling open source editing, being licensed under LGPL.
Above all, the Fast Artificial Neural Network Library continues to rank among the most reliable and efficient neural network software solutions globally. With support available for various programming languages, including PHP, Python, Delphi, and Mathematica bindings, this software remains a top choice among developers globally.
Version 2.1.0 Beta: N/A