Skip to content

andrewahn-ubc/feed-forward-neural-network

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Feed-Forward Neural Network

A simple neural network built from scratch (ie. using only numpy, no tensorflow or pytorch) as an exercise to learn machine learning.

Background

I'm a complete ML noob. There are a lot of tutorials on how to build an NN from scratch but I want to figure it out on my own so I'll probably stumble around make dumb mistakes. I watched 3B1B's series on neural networks to get a theoretical understanding of NNs, but other than that, I'm going in blind. That's my goat btw

"What I cannot create, I do not understand" -Richard Feynman

Implementation Details

I'll flesh out this section with what design choices I made once I've finished building it lol

Some goals

  • Once the implementation is done, test it on the MNIST hand-written digits dataset and get a relatively high accuracy (anything >80%).
  • Explore some fun deep learning techniques (learning rate schedules, early stopping, different gradient descent variations, hyperparameter tuning, different initializtion methods, etc.).