Hand Digit Neural Network ML

The goal of this small project is to implement a simple neural network learning algorithm to be able to recognize digits drawn by a user on a 50x50 canvas.

Temporarily hosted on: http://ml.vjaouen.com/

How to launch locally

Prerequirements:

• Nodejs (https://nodejs.org/en/)
• Yarn (https://yarnpkg.com/)
• PostgreSQL

Install and run

• Clone repository locally
• Run yarn install
• Create digitml database and run migrations with yarn knex migrate:latest --knexfile ./knexfile.js
• Run yarn start
• Open http://localhost

Learn

Just run yarn learn.

It will iterate on lambdas from 1 to 10. Change the parameter LEARN_CHUNK to the ratio of learning data/cross-validation data as desired.

Precision

Run yarn compute-precision