Homework 2

A neural network for visually classifying points in a two dimensional input space into two classes, posted because the visualization may be useful.

A precompiled version is available in the downloads section.

The input files are in the format "X\tY\tTARGET" where X and Y are floating point values between -10 and 10, and TARGET is 0 or 1.

There is no testing set input. The only output of this program is the error at each epoch and a graphical view of the network.

main.c is responsible for the rendering loop.

shaderbuilder.c builds a fragment shader that classifies each point based on it's texture coordinate and the weights array.

nn.c builds and trains an mlp or rbf.

Usage is "homework2 [-f <file.dat>] [-r ] ". The arguments must be provided in this order!

I didn't bother with much error handling so this will probably crash if something goes wrong. Works fine for me on my computer so long as the specified file exists and I don't do anything like make an rbf without an rbf layer.