An Early Attempt at a Neural Network


Put numbers in the boxes below in order to get started.

Input goes here

These inputs will be fed into the left side of this network. The top input field corresponds to the top neuron. The inputs will then be processed by the neurons in the middle, before being shot out the other side as the output.



This is the answer that the network should come up with.

Test Batch Size

Test size is the amount of testing the network will do to determine whether or not it has made a good choice. Bigger sizes will make it slower and more accurate. Smaller sizes will make it faster but more likely to make an error. "Auto" will make it automatically try 100 colors.


Right / Wrong

Whether it got the puzzle right or wrong.

Output Log

The actual outputs from each neuron.

This graph shows the networks success rate over time.

My first attempt at machine learning. Backpropagation is the bigest issue here but technically it does learn.

Mar 16, 2017