Go to JKU Homepage
institute-for-application-oriented-knowledge-processing
What's that?

Institutes, schools, other departments, and programs create their own web content and menus.

To help you better navigate the site, see here where you are at the moment.

Interpreting Chess Evaluation Functions

Supervisor: Johannes Fürnkranz

The goal of this thesis is to apply novel techniques for interpreting black-box models, such as LIME [1], to learned (and possibly also hand-coded) chess evaluation functions, such as those that are currently in used in programs such as Leela chess 0, and to assess inasmuch they can be used to gain insight into the playing behavior and/or to characterize the playing style of different chess playing entities. Similar techniques could, e.g., also be used to assess differences in evaluation functions learned from human players of different strength (e.g., amateuer vs. grandmaster) or of different stype (e.g., Tal vs. Capablanca).

If desired, a similar topic can also be defined for games or puzzles other than chess.

Literature

[1] Marco Túlio Ribeiro, Sameer Singh, Carlos Guestrin: "Why Should I Trust You?": Explaining the Predictions of Any Classifier, opens an external URL in a new window. KDD 2016: 1135-1144.