Case-based explanations - A small demo

About

Case-based explanations help understand the rational behind Machine-learning classifiers through the use of historical examples.
This demo showcases a simple case-based explanation system for a handwritten digit classifier. The system will evaluate a user-submitted drawing and present a set of similar examples based on the features computed by a Neural Network. For more information visit our About page.

Instructions

  1. Get started by drawing a number from 0 to 9.
  2. Submit your drawing by clicking the button.
  3. Wait for our API to process your drawing.
  4. See the results below.