https://github.com/sharpcoder/protonet
protonet is an implementation of the backpropagation algorithm written in C#
https://github.com/sharpcoder/protonet
Last synced: 9 months ago
JSON representation
protonet is an implementation of the backpropagation algorithm written in C#
- Host: GitHub
- URL: https://github.com/sharpcoder/protonet
- Owner: SharpCoder
- License: mit
- Created: 2013-10-04T15:56:25.000Z (over 12 years ago)
- Default Branch: master
- Last Pushed: 2014-07-01T01:56:15.000Z (over 11 years ago)
- Last Synced: 2025-03-28T21:21:45.430Z (10 months ago)
- Language: C#
- Homepage:
- Size: 188 KB
- Stars: 7
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
Proto-Net
========
This project is a barebonese Neural Network designed to be an example of how to implement backpropagation.
My goal was to heavily comment the code and make it as logical as possible. There is a fair amount of recursion
and, although it is not the most efficient implementation, I wanted to make this neural network easier
to digest from a comprehensive perspective and focused on trying to point out what each thing does.
I plan to port my code to a few different languages. I've tried to keep most of the comments intact.
FastNet
========
There is a class included which is called "FastNet" and it represents an incredibly small, array-based neural network
implementation. It has a hard-coded topology, but is very interesting if you want to see the iterative steps that
are taken. Additionally, it's incredibly fast (compared to the recursive based approach).
Background
========
After spending an ungodly amount of time trying to make sense of the backpropagation algorithm,
something clicked for me. I wish to share this knowledge with others, hopefully saving someone a year of pondering!