Home Page

Manual

- overview

- data structures

- preprocessing

- networks

- classification

- graphing

- examples

 

NetMaker is a freeware package for building neural networks. It was developed primarily to support particle interaction classification in the high energy physics experiments at CERN - SMC, COMPASS, and the ICARUS, a neutrino experiment at LNGS. It was also my test-bed for experiments with various algorithms (some well known and some of my own ideas). Now it turned into a 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. >>

The project development was suspended for a really long time. Now the neural network topic becomes hot again... We are happy with the performance of NetMaker engine, but it absolutely needs modernization on the interface! It will come - as a part of the new, simple programming language POOL.

neural engine features
papers

Download current version:  v0.9.5.2

Bayesian Framework features are available now, as a result of our latest work on EM form factor parametrizations.

- history of changes

previous versions:

v0.9.4.3
v0.9.3.4
v0.9.2.8
v0.9.2.0

ROOT interface with an example.
(ROOT homepage)

ROOT-based C++ code to run the trained network in your programs.

The page is translated to Serbo-Croatian language by Web Geeks.

Shall you have any questions, comments - e-mail me.

NetMaker window

• clear visualization of input data and network results:
      · specialized plots - purity-efficiency, background rejection,
      · histograms - with μ and σ of the distribution,
      · scatter plots - with Rxy linear correlation coefficient,
      · training parameters vs iteration,
      · real-time updated plots;
• written in C#;
• calculations optimized in C/asm (SSE2/SSE3/SSE4.1 aware).