Ti-Rong Wu

LaGO: Latent Action Guidance for Online Reinforcement Learning

Kuan-Yen Liu, Ren-Jyun Huang, Ti-Rong Wu

Evaluating Game Difficulty in Tetris Block Puzzle

Chun-Jui Wang, Jian-Ting Guo, Hung Guei, et al.

This paper uses an AI agent called Stochastic Gumbel AlphaZero to evaluate how difficult different versions of Tetris Block Puzzle are by measuring how quickly the AI learns and performs. The researchers tested various rule changes—like allowing players to hold blocks or preview them—and found that these features make the game easier, while adding more block types makes it harder. This approach provides game designers with a principled way to assess and balance puzzle game difficulty.