Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/danesherbs/hopfieldnetwork
https://github.com/danesherbs/hopfieldnetwork
Last synced: 3 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/danesherbs/hopfieldnetwork
- Owner: danesherbs
- Created: 2015-10-23T18:20:19.000Z (about 9 years ago)
- Default Branch: master
- Last Pushed: 2015-10-23T22:15:38.000Z (about 9 years ago)
- Last Synced: 2024-11-09T16:47:41.406Z (about 1 month ago)
- Language: Python
- Size: 160 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Hopfield Network
Hopfield networks remember given configurations, or "patterns". It does so by following the Hebbian rule for learning:
##### "Neurons that fire together, wire together. Neurons that are out of sync, fail to link."
Patterns are a sequence of states (which only take the values +1 and -1). For example, a pattern may be [1,-1,1], but couldn't be [2,1,-4], since it contains numbers other than +1 and -1.
For this Hopfield network, you can pass a pattern of any size
python hopfield.py 1 -1 -1 1 1 -1
A random sequence of states will be generated. The network's job is to get from this sequence of states **to your pattern** (without looking at it -- *that would be cheating*!)
The network will then update each node according to simple rules, and arrive at your pattern.
```
RANDOM INITIAL STATE
[1, 1, -1, 1, 1, -1]
UPDATE NODE 0
Weights: [ 0 -1 -1 1 1 -1]
Inputs: [1, 1, -1, 1, 1, -1]
Field: 1
UPDATE NODE 1
Weights: [-1 0 1 -1 -1 1]
Inputs: [1, 1, -1, 1, 1, -1]
Field: -1
UPDATE NODE 2
Weights: [-1 1 0 -1 -1 1]
Inputs: [1, -1, -1, 1, 1, -1]
Field: -1
UPDATE NODE 3
Weights: [ 1 -1 -1 0 1 -1]
Inputs: [1, -1, -1, 1, 1, -1]
Field: 1
UPDATE NODE 4
Weights: [ 1 -1 -1 1 0 -1]
Inputs: [1, -1, -1, 1, 1, -1]
Field: 1
UPDATE NODE 5
Weights: [-1 1 1 -1 -1 0]
Inputs: [1, -1, -1, 1, 1, -1]
Field: -1
END STATE: [1, -1, -1, 1, 1, -1]
END STATE
[1, -1, -1, 1, 1, -1]
```
Note that Hopfield networks are **sign blind** i.e. if [1, -1, 1] is an attracting fixed point, [-1, 1, -1] is too.