JAM Networks

Home made Java & Matlab Neural Networks, made by Mark

Main
The predictions of a Neural Network on the UNIPEN dataset:

The thickness represents the confidence of the network

This browser does not have a Java Plug-in.
Get the latest Java Plug-in here.

Enter and Backspace:Browse
plus or minus : Larger or smaller digits
< or > : More or less digits
[ or ] : Thinner or thicker digits.

C: Toggle view only misclassifications and view everything.

JAM Neural Networks is a software package for experimenting with Neural Networks:
  • BackPropagation
  • Resilient BackPropagation
  • Perceptrons with the delta rule
  • AdaBoostM2 ensemble
All these algorithms are implemented very straighforward in Java - so they are easy to understand, modify and extent. However, for analyzing and running these Networks, we use MatLab. With MatLab, you can easily write experiments, plot weight-changes, etc.

Last modified on Thu Oct 29 18:01:27 2009.
This page uses google analytics and this work is licenced under: "Creative
WWW.BOLTWIRE.COM