Matchbox Educable Noughts and Crosses Engine
term_id: matchbox_educable_noughts_and_crosses_engine
Category: basic_concepts
Definition
The ME-Noughts-and-Crosses Engine was an early demonstration of machine learning, specifically reinforcement learning. Constructed from 304 matchboxes, each representing a unique board state, the system used colored beads to represent possible moves. After playing against a human opponent, the operator would reinforce successful moves by adding beads and remove beads from losing paths. Over time, the machine learned optimal strategies through trial and error, serving as a tangible precursor to modern AI algorithms like Q-learning.
Summary
A physical reinforcement learning device built by Donald Michie in 1961 to play Tic-Tac-Toe using matchboxes and beads.
Key Concepts
- Reinforcement learning
- Trial and error
- Historical AI
- Tic-Tac-Toe
Use Cases
- Educational demonstrations of RL
- History of AI studies
- Conceptualizing reward-based learning