This is a simple model, that tries to approximate the function y = m * x + b. For the slope(m) I choosed 1 and the y-intercept(b) is 0. Under Visualisation you can find some sample datapoints, and the guessed function at the left, that updates in real-time. At the right is the gradient landscape where try to go to the dark blue area where m is 1 and b is 0.
This repository contains the logic and scripts that combine three packages:
- TensorFlow.js, libary to process neural network operations in JavaScript
- visjs.org, a libary to show 3dGraphs
- Babeljs.io, an extention to use next generation JavaScript
- Parcel Bundler, Blazing fast, zero configuration web application bundler
- Google Charts, Display live data on your site
Visualisation of the function we want to approximate:
Visualisation of the gradient Landscape:
- Download the project
- Call 'npm install' in your projectfolder(Make sure npm is installed)
- 'npm run build'
- 'npm run watch'
- If your browser doesn't open by itself navigate to 'http://localhost:1234/'