Java neural network framework Neuroph released version 2.4, which brings important improvements related to learning rules, performance improvements, and new features in GUI and API.
Probably one of the most important new features is the additional settings and performance improvements for backpropagation learning rule, which significantly improves learning results of neural networks that use it. This also improves performance in all tools that use it, like the image recognition tool.
Some new application related features include basic character and handwritting recognition support and some stock market prediction samples. Also, there are many improvements in API an d GUI which make Neuroph more intuitive and easier to use.
In this release Neuroph switched licensing from LGPL3 to Apache2, after the initiative for integration with Mahout (the Apache machine learning library), and acceptance of the GSoC project proposal. The new licencing provides more freedom to potential commercial users and enables the broader adoption of the framework.
Porting to the NetBeans Platform
After the release of 2.4, development focus will be on porting all the GUI to the NetBeans Platform in order to provide a state of the art neural network editor.
The porting project has allready made nice progress and previews are publicly available from SVN. Here is a screenshot (click to enlarge it):
Follow our porting progress here!