https://github.com/ruffrey/nurtrace

Neural network library implementing Hebbian plasticity with rebalancing of N-ary trees and node growth
https://github.com/ruffrey/nurtrace

Last synced: over 1 year ago
JSON representation

Neural network library implementing Hebbian plasticity with rebalancing of N-ary trees and node growth

• Host: GitHub
• URL: https://github.com/ruffrey/nurtrace
• Owner: ruffrey
• Created: 2017-02-05T16:19:29.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2018-01-01T19:48:30.000Z (over 8 years ago)
• Last Synced: 2025-01-08T18:50:12.685Z (over 1 year ago)
• Language: Go
• Homepage:
• Size: 1010 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 14
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Name Pending (codename nurtrace)

v0.12.0

This is a real neural network, modeled after real neuring firing patterns.

It uses the concept of the Action Potential and Membrane Potential from neuroscience to

train a neural network.

## Todos and Project Management

See the github issues for this project.

## Core lib architected to support:

- [x] simple math resembling millivolts in neural network firings

- [x] fast and works cross platform

- [x] can train in parallel on the same dataset

- [x] parallelized learning and inputs

      - instead of only training on one thing at a time, can train on more than one stimuli simultaneously

- [x] focus is on the pathway, rather than the neural network solving only one problem

- [x] generalized model can be used for any kind of data that can be put in memory

- [x] same network can be trained on more than one kind of stimuli to solve different problems

- [ ] methods for adaptive learning - training evaluates its learning speed and adjusts

- [ ] (maybe) parallelize and use [SIMD](https://github.com/bjwbell/gensimd) instructions

## Stretch Goals

- [x] Provides tools for easy distributed training across computers

- [x] Visualize network changing over time

    - [ ] visualize with built-in tools

- [ ] Prebuilt network training for:

    - [ ] character and word recurrent neural networks

    - [ ] image classification

    - [ ] binary decision

- [ ] Tools for measuring training speed

- [ ] Cross platform GUI app for all of it