diwasx / neural_network_scratch

Basic neural network library

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

neural_network_scratch

Basic Neural Network Library from scratch.

Requirement

python3, pip3

Installation

git clone https://github.com/diwasx/neural_network_scratch
cd neural_network_scratch
pip3 install -r requirements.txt --user

Library usage

Creating object and initializing constructor:

NeuralNetwork (inputLength, hiddenLength, outputLength)

n = NeuralNetwork(3, 10, 9)

Generating output using FeedForward

n.feedForward(inputs)

Neural Network Structure

n.nnStructure()

Training NN with know data (supervised learning)

n.trainSVLearing(inputs,targets,learningRate)

Training Visualization with weights and biases changes

n.trainSVLearingVisualization(inputs,targets,learningRate)

Library Implementation (Examples)

  • Logical Gates
  • Digit Recognition
  • Machine Play T-rex game using NN and genetic algorithm (NEAT algorithm)

About

Basic neural network library

License:MIT License


Languages

Language:Python 100.0%