https://github.com/hetelek/Neural-Network-Playground

A neural network Swift playground, with no third party dependencies.
https://github.com/hetelek/Neural-Network-Playground

computer-vision neural-network swift-playgrounds wwdc-scholarship

Last synced: 25 days ago
JSON representation

A neural network Swift playground, with no third party dependencies.

• Host: GitHub
• URL: https://github.com/hetelek/Neural-Network-Playground
• Owner: hetelek
• Created: 2017-03-13T20:52:19.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2019-02-12T06:49:50.000Z (almost 6 years ago)
• Last Synced: 2024-04-22T12:30:57.503Z (8 months ago)
• Topics: computer-vision, neural-network, swift-playgrounds, wwdc-scholarship
• Language: Swift
• Homepage:
• Size: 3.18 MB
• Stars: 378
• Watchers: 12
• Forks: 6
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• Awesome-Swift-Playgrounds - Neural Network Playground - A neural network Swift playground, with no third party dependencies. (PlaygroundBooks)

README

        
# Neural Network Playground

This project uses iOS Playgrounds to display a working neural network. No external libraries are used. The [matrix](https://github.com/hetelek/Neural-Network-Playground/blob/master/Neural%20Network.playground/Sources/Matrix.swift) and [neural network](https://github.com/hetelek/Neural-Network-Playground/blob/master/Neural%20Network.playground/Sources/Network.swift) code is written in pure Swift. This playground has only been tested on an 12.9‑inch iPad Pro.

### Device Installation

This is how I do it, but I'm sure there is a better way:

1. Install the [Swift Playgrounds app](https://itunes.apple.com/us/app/swift-playgrounds/id908519492?mt=8) on your device.

2. Download or clone this repository.

3. AirDrop the `Neural Network.playground` folder from your Mac to your device.

4. Open the received playground in the Playground application.

### Screenshots

Here is the page that predicts a drawn digit:

![Custom Digit](images/custom.png?raw=true "Custom Digit")

And a neural network being trained to fit some points:

![Fitting](images/fitting.png?raw=true "Fitting")