This is a simple demo of binary classification, in which the machine tries to recognize patterns drawn by hand. It shows how one can build a training set by labeling examples and then make predictions for pictures of unknown patterns.
This demo implements two classifiers: $K$-nearest neighbors and the perceptron.
Start by drawing a pattern and .
Then, draw a different pattern and .
Do this multiple times to collect a sample of drawings (of these 2 different patterns).
When you have enough data, draw a new example and ask the machine to
or .
At any time, you can or