Learning / AI
Learning Maze
An interactive maze for exploring how reinforcement learning changes the odds, and how training differs from inference.
An interactive reinforcement-learning explainer. Send explorers through a maze and watch their results change what the doors remember.
Compare training with inference, switch between relative weights and percentages, and watch the chance of reaching the birthday cake change with practice. The door analogy illustrates learning; it is not a literal neural network or conventional backpropagation.
The simulation runs entirely in your browser, with no account or backend required.