https://github.com/ericjang/rnn-dynamics

Code and report for APMA136 Final Project
https://github.com/ericjang/rnn-dynamics

Last synced: 5 months ago
JSON representation

Code and report for APMA136 Final Project

• Host: GitHub
• URL: https://github.com/ericjang/rnn-dynamics
• Owner: ericjang
• License: mit
• Created: 2015-05-04T17:45:37.000Z (about 11 years ago)
• Default Branch: master
• Last Pushed: 2015-05-06T05:44:09.000Z (about 11 years ago)
• Last Synced: 2024-12-26T20:42:56.842Z (over 1 year ago)
• Language: Matlab
• Size: 1.2 MB
• Stars: 19
• Watchers: 4
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# RNN-dynamics

Code for my APMA136 Final Project, based on "Opening the Black Box : Low-Dimensional Dynamics in High-Dimensional Recurrent Neural Networks".

Codes for `simple2d`, `threebit_flipflop`, `motor_control` tasks are found in respective subdirectories. All code written by Eric Jang, under the MIT license. Most of the code is implemented in MATLAB, although there is a bit of python here and there.

## Trained Submanifold Attractor

3-bit Flip Flop

![Imgur!](http://i.imgur.com/4bBZevB.gif)

4-bit Flip-Flop

![Imgur!](http://i.imgur.com/umIvYsU.gif)

4-bit Flip-Flop, Logistic Nonlinearity

![Imgur!](http://i.imgur.com/EwYzp9n.gif)

4-bit Flip-Flop, pre-trained on 3-bit FF

![Imgur!](http://i.imgur.com/GRZyFwO.gif)

## Formation of Phase Space from Training

3-bit FF

![Imgur!](http://i.imgur.com/S2LXC7i.gif)

4-bit FF

![Imgur!](http://i.imgur.com/QyMiRGL.gif)

Formal, academic writeup and informal blog posts are coming soon.