Skip to content

Latest commit

 

History

History
10 lines (10 loc) · 596 Bytes

README.md

File metadata and controls

10 lines (10 loc) · 596 Bytes

Neural Network from scratch

Created for Machine Learning class at WPI.
Implemented a neural network from scratch that trains on the mnist data set.
Created a hyperparamter optimization algorithm that optimizes each hyperparamter individually and combines then together.

Achieves 95% accuracy after 1 epoch, and 97% after 30 epochs.

Visualization of the training using PCA.

Blue surface is the training waits from an inverse transform of 2 value PCA.
Red dots are the approximated neural network weights array.
The Z dimension is the cross entropy loss. image