NetMaker is the open source system for building the neural networks. It was developed primarily to support particle interaction classification in high energy and neutrino physics experiments at CERN (SMC, COMPASS, ICARUS). It was also my test-bed for experiments with various algorithms (some well known and some of my own ideas). Now it turned into quite a flexible network designer with many possible applications - the idea is to have networks, pre-/post-processing and data sets split into uniformly working, task independent blocks which may form bigger systems.  >>

→ neural engine features
→ papers

Current version is: v.15.02.2008b

   - download, license, installation notes
   - sources: on request (e-mail me)
   - history of changes

Next major steps:
   - image processing support;
   - communication interface between NetMaker and other applications.

previous versions:

v.02.09.2007b
v.29.06.2007e
v.21.10.2006d

Framework.NET 2.0 or 3.0 - necessary to run the program under Windows.

ROOT interface with an example. (ROOT homepage)

C++ code to run the network in your programs. (updated ages ago)

Shall you have any questions, comments - e-mail me: Robert Sulej

modified on 24-02-2008
NetMaker window


◦ optimized for speed while maintaining huge data sets;
◦ clear visualization of input data and network results:
      ∙ histograms - with m and s of the distribution,
      ∙ scatter plots - with Rxy linear correlation coefficient,
      ∙ error vs training iteration, real-time updated plots,
      ∙ purity-efficiency plots - specialized for classification;
◦ written in C#;
◦ calculations optimized in C/asm (SSE2/SSE3 aware).