Neural net 1-d regression demo using convnet.js

A neural net is trained to predict a scalar output value (y) for a given scalar input value (x).
The neural network architecture is specified in the text box below.
Read the convnetjs documentation for info on how to modify the network and the trainer.

This version is adapted for MCB 419. Original version by Andrej Karpathy is here.

# training points:

click checkbox to also show outputs of a particular layer (use buttons below to select layer)

javascript code will add layer-select buttons here