Java Neural Network Framework Neuroph / News: Recent posts

Java Neural Network Framework Neuroph 2.4 has been released. This version brings new features for training neural networks, significant performance improvements, some sample OCR tools and API, and learning vizualisation samples.

Version 2.3.1 of Java Neural Network Framework Neuroph which brings some important bug fixes and improvments has been released. This release fixes reported bugs for image recognition, LMS learning formula, and bugs when creating large networks. Graph view has been migrated to JUNG 2, and specific network layouts are added.

New version brings some important features which wil make neuroph easier to use for practical tasks, and also to experiment with. New features include: momentum for backpropagation, importing training sets from txt files, error graph, xml support, basic help system and more.

The version 2.1.1b of Java neural network framework Neuroph has been released. This version brings few bug fixes and new features. They include unsupervised hebbian learning and some gui improvments.

The version 2.1.0b of Java neural network framework Neuroph has been released. This version brings cleaned and optimized code, full api documentation, Java neural network library, improved GUI network editor, and plugin architecture.

The version 2.0.0 alpha of neural network feamework neuroph has been released. Since there were many big changes on previous version 1.0.0a, we decided to switch to version 2. This version brings support for new neural architectures, competitive networks, serialization and new GUI application. This is now the recommended realase.
For the users of previous versions there may be compativbility issues, but they can be resolved relatively easily. Still aiming to be out of the box easy to use framework.

The version 1.0.1 of framework has been released. This version bring serialization support to version 1.0.0 and many optimization improvments