Build your own MENACE (matchbox-powered machine learning computer)

mscroggs.co.uk

38 pointsJNRowe3 comments

Two years ago I used MENACE to introduce students to machine learning approaches: it was the final example in course of programming for lawyers.

https://pypi.org/project/lex-game/

The code uses excel files instead of matchboxes... It was in fact an example intended to explore further the use of some libraries for office automation. It also allowed for easy seeing of the weights for different choices, still in a structured way even if outside the program itself.

What would this algorithm be in modern day machine learning terms? Is it similar to anything we have now?

I think it would be most similar to tabular Q-learning. It has some weaknesses due to its simplicity, like not decaying the learning rate (you always add/remove from a fixed number of matches; maybe if you used an unbounded number of matches it'd be equivalent?) or having any exploration built-in, so it doesn't enjoy the guarantees tabular Q-learning does like converging to the optimal game play - MENACE would presumably jitter around the optimal strategy if it approached it at all.

Reply on news.ycombinator.com