Solving the Hole in the Square Problem with a Neural Network

the hole

error

visualize the net

reset the net

try the net

generate vectors

learn data

load a trained net

hidden level 1

4

hidden level 2

4

hidden level 3

4

learning rate

0.3

error threshold

0.02

learning ages

1000

vectors for learning

50

radius

0.7

vectors for testing

300

This Demonstration shows a neural network with backpropagation error. The net is tested with the problem of "the hole in a square", which means that given the coordinates of a random point, the neural network should determine whether the point is in the hole or not.