https://github.com/ruffrey/dynamic-middle-out
dynamical neural network with novel features loosely similar to reservoir computing
https://github.com/ruffrey/dynamic-middle-out
Last synced: 12 months ago
JSON representation
dynamical neural network with novel features loosely similar to reservoir computing
- Host: GitHub
- URL: https://github.com/ruffrey/dynamic-middle-out
- Owner: ruffrey
- Created: 2017-07-15T13:42:21.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2017-07-19T02:52:54.000Z (almost 9 years ago)
- Last Synced: 2025-02-26T05:26:45.171Z (over 1 year ago)
- Language: Go
- Size: 8.79 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Dynamic Middle-Out reservoir neural network
A custom reservoir-like neural network based off my understanding
of neuroscience and vague reading of high level papers on liquid
state machines (though they are few and far between). May not
be very similar as the math is beyond me at this point, but the
goal is not to be a traditional reservoir computing network.
Key differences as I understand them:
- use of tiny int8 for speed, instead of floats, which fundamentally
is different from other neural networks
- more parts of the system are dynamic (?)