Classification Explorer

An interactive explorer for two of the simplest machine-learning models — k-nearest-neighbour classification and a small neural-network regression — built to show, visually, what a model has actually learned from a handful of examples.

Hold a number key from 1 to 9 and move the mouse over the left canvas to lay down points labelled with that class. The classifier retrains on every addition and colours the whole plane by its prediction, so decision boundaries appear and shift as data is added. A regression model trained on the same points reports its continuous output for the cursor position and the delta from the classifier's answer; Sample renders that regression as a colour field on the right-hand canvas.

Built on RapidLib, the interactive machine-learning library behind the MIMIC project, as a teaching aid — a way to make the building blocks of AI legible to students before they applied them to their own practice.