https://github.com/anuraganalog/neural-networks

Implementing Neural Networks from scratch
https://github.com/anuraganalog/neural-networks

backpropagation dataset feedforward hopfield-network metrics neural-networks neurons programs python3 splitting testing training visualization

Last synced: 2 days ago
JSON representation

Implementing Neural Networks from scratch

• Host: GitHub
• URL: https://github.com/anuraganalog/neural-networks
• Owner: AnuragAnalog
• License: gpl-3.0
• Created: 2020-03-07T11:12:52.000Z (almost 5 years ago)
• Default Branch: master
• Last Pushed: 2022-10-15T19:22:28.000Z (about 2 years ago)
• Last Synced: 2024-11-18T07:41:43.796Z (about 1 month ago)
• Topics: backpropagation, dataset, feedforward, hopfield-network, metrics, neural-networks, neurons, programs, python3, splitting, testing, training, visualization
• Language: Python
• Size: 203 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Neural Networks

My implementation of different neural network models from scratch in Python

## NN Models

* Single Layer Perceptron

* HopField Network

## Activation Functions

It's part in the **Nerual Networks**, which says whether the neuron should be active or not.

* *Sigmoid*

It is a characteristic S-shaped function; the domain of the function is R, which maps to a range of [0, 1], also called a logistic function.

![sigmoid](./Activation-functions/sigmoid.png)

* *ReLU*

It is a piecewise function, which outputs the number itself if +ve else 0.

![relu](./Activation-functions/relu.png)

* Softmax

It's a mathematical function which gives a normalized exponential vector of n-input vector, which is probabilistic distribution.

![softmax](./Activation-functions/softmax.jpg)

## Assignments

> Check the readme's in the directories for more description

* assignment1: Implementing Single layer Perception

* assignment2: Implementing Hopfield network