A Continuous Information Gain Measure to Find the Most Discriminatory Problems for AI Benchmarking
This provides a more efficient benchmarking approach for AI researchers, though it is incremental as it applies existing information theory to a specific domain.
The paper tackles the problem of efficiently benchmarking AI algorithms by introducing an information-theoretic method to select a minimal subset of problems that maximally discriminates between agents, tested on the General Video Game AI framework where a handful of games achieve nearly the same discriminatory accuracy as over a hundred games.
This paper introduces an information-theoretic method for selecting a subset of problems which gives the most information about a group of problem-solving algorithms. This method was tested on the games in the General Video Game AI (GVGAI) framework, allowing us to identify a smaller set of games that still gives a large amount of information about the abilities of different game-playing agents. This approach can be used to make agent testing more efficient. We can achieve almost as good discriminatory accuracy when testing on only a handful of games as when testing on more than a hundred games, something which is often computationally infeasible. Furthermore, this method can be extended to study the dimensions of the effective variance in game design between these games, allowing us to identify which games differentiate between agents in the most complementary ways.