Hung Guei 桂浤

Highlights

• Proposed OptionZero, which integrates options into MuZero with autonomous option discovery during training, achieving an average score of 131.58% on the Atari benchmark. Accepted in ICLR 2025 with an oral presentation (1.8% of 11,500 papers).
• Investigated the interpretability of MuZero and demonstrated that MuZero can maintain its planning performance by correcting the errors in the dynamics network. Published in IEEE Transactions on Artificial Intelligence and TAAI 2024 (Runner-Up Paper Award among 54 accepted papers).
• Developed an AlphaZero-based solving method with online fine-tuning. Experiments in 7x7 Killall-Go showed that the same tasks could be solved using 24% of the computational resources required by the baseline. Related works published in NeurIPS 2023, ACG 2025, and IEEE Transactions on Games.
• Developed MiniZero, a general framework for AlphaZero, MuZero, Gumbel AlphaZero, Gumbel MuZero, and other zero-knowledge reinforcement learning methods. Published in IEEE Transactions on Games and open-sourced on GitHub (received 100+ stars).
• Studied the strength adjustment method for Monte Carlo tree search. Achieved near-linear strength adjustment by dynamically adjusting the move selection during the search. Published in IEEE Computational Intelligence Magazine, AAAI-19, and two patents.
• Developed a 2048 game-playing program using RL methods, achieving an average score of 625,377 points and a 72% rate for reaching the 32768-tile. Published in IEEE Transactions on Games and recognized as the state-of-the-art RL program for 2048 as of September 2021.

Publications

1. Qian-Rong Li, Hung Guei, I-Chen Wu, Ti-Rong Wu, “MAPLE: Multi-State Aggregated Policy Evaluation for AlphaZero in Imperfect-Information Games,” IEEE Conference on Games (IEEE CoG 2026), Madrid, Spain, Sep. 2026.
2. Hung Guei, Yan-Ru Ju, Wei-Yu Chen, Ti-Rong Wu, “Demystifying MuZero Planning: Interpreting the Learned Model,” IEEE Transactions on Artificial Intelligence, vol. 7, no. 2, pp. 1025–1036, Feb. 2026.
3. Chung-Chin Shih, Ti-Rong Wu, Ting Han Wei, Yu-Shan Hsu, Hung Guei, I-Chen Wu, “A Study of Solving Life-and-Death Problems in Go Using Relevance-Zone Based Solvers,” IEEE Transactions on Games, in press, Dec. 2025.
4. Chun-Jui Wang, Jian-Ting Guo, Hung Guei, Chung-Chin Shih, Ti-Rong Wu, I-Chen Wu, “Evaluating Game Difficulty in Tetris Block Puzzle,” The 30th Game Programming Workshop (GPW-25), Kanagawa, Japan, Nov. 2025.
5. Chi-Huang Lin, Ting Han Wei, Chun-Jui Wang, Hung Guei, Chung-Chin Shih, Yun-Jui Tsai, I-Chen Wu, Ti-Rong Wu, “Relevance-Zone Reduction in Game Solving,” Advances in Computer Games 2025 (ACG 2025), online, Oct. 2025.
6. Po-Wei Huang, Pei-Chiun Peng, Hung Guei, Ti-Rong Wu, “OptionZero: Planning with Learned Options,” The Thirteenth International Conference on Learning Representations (ICLR 2025), Singapore, Apr. 2025. (Oral presentation: 1.8%, acceptance rate: 32.08% among 11,500 papers)
7. Ti-Rong Wu, Hung Guei, Pei-Chiun Peng, Po-Wei Huang, Ting Han Wei, Chung-Chin Shih, Yun-Jui Tsai, “MiniZero: Comparative Analysis of AlphaZero and MuZero on Go, Othello, and Atari Games,” IEEE Transactions on Games, vol. 17, no. 1, pp. 125–137, Mar. 2025.
8. Hung Guei, Yan-Ru Ju, Wei-Yu Chen, Ti-Rong Wu, “Interpreting the Learned Model in MuZero Planning,” The 29th International Conference on Technologies and Applications of Artificial Intelligence (TAAI 2024), Hsinchu, Taiwan, Dec. 2024. (Runner-up paper award among 54 accepted papers)
9. Yun-Jui Tsai, Ting Han Wei, Chi-Huang Lin, Chung-Chin Shih, Hung Guei, I-Chen Wu, Ti-Rong Wu, “Solving 7x7 Killall-Go with Seki Database,” The Computers and Games 2024 Conference (CG 2024), online, Nov. 2024.
10. Po-Ting Chen, Chien-Liang Kuo, De-Rong Sung, Hung Guei, I-Chen Wu, “MuMu Won the EinStein Würfelt Nicht! Tournament,” ICGA Journal, vol. 45, no. 2–3, pp. 81–84, Jan. 2024.
11. Ti-Rong Wu*, Hung Guei*, Ting Han Wei, Chung-Chin Shih, Jui-Te Chin, I-Chen Wu, “Game Solving with Online Fine-Tuning,” The Thirty-Seventh Annual Conference on Neural Information Processing Systems (NeurIPS 2023), New Orleans, USA, Dec. 2023. (Acceptance rate: 26.1% among 12,343 papers; * contributed equally)
12. Chien-Liang Kuo, Po-Ting Chen, Hung Guei, De-Rong Sung, Chu-Hsuan Hsueh, Ti-Rong Wu, I-Chen Wu, “An Empirical Analysis of Gumbel MuZero on Stochastic and Deterministic Einstein Würfelt Nicht!,” The 2023 Conference on Technologies and Applications of Artificial Intelligence (TAAI 2023), Yunlin, Taiwan, Dec. 2023.
13. Hung Guei, “On Reinforcement Learning for the Game of 2048,” Ph.D. Dissertation, Institute of Computer Science and Engineering, National Yang Ming Chiao Tung University, Hsinchu, Taiwan, Jan. 2023. (Three Best Ph.D. Thesis Awards and one Honorable Mention Award)
14. Chih-Yu Kao, Hung Guei, Ti-Rong Wu, I-Chen Wu, “Gumbel MuZero for the Game of 2048,” The 2022 Conference on Technologies and Applications of Artificial Intelligence (TAAI 2022), Tainan, Taiwan, Dec. 2022.
15. Hung Guei, Lung-Pin Chen, I-Chen Wu, “Optimistic Temporal Difference Learning for 2048,” IEEE Transactions on Games, vol. 14, no. 3, pp. 478–487, Sep. 2022.
16. I-Chen Wu, Ti-Rong Wu, An-Jen Liu, Hung Guei, Ting-Han Wei, “Method for adjusting the strength of turn-based game automatically,” U.S. Patent, US11247128B2, Feb. 2022.
17. I-Chen Wu, Ti-Rong Wu, An-Jen Liu, Hung Guei, Ting-Han Wei, “Method for automatically modifying strength of turn based game,” R.O.C. Patent, TWI725662B, Apr. 2021.
18. An-Jen Liu, Ti-Rong Wu, I-Chen Wu, Hung Guei, Ting-Han Wei, “Strength Adjustment and Assessment for MCTS-Based Programs,” IEEE Computational Intelligence Magazine, vol. 15, no. 3, pp. 60–73, Aug. 2020. (Impact factor 11.356)
19. Hung Guei*, Ting-Han Wei*, I-Chen Wu, “2048-like Games for Teaching Reinforcement Learning,” ICGA Journal, vol. 42, no. 1, pp. 14–37, May 2020. (* contributed equally)
20. Hung Guei, Ting-Han Wei, I-Chen Wu, “Teaching Reinforcement Learning and Computer Games with 2048-Like Games,” The 33rd Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2019), Niigata, Japan, Jun. 2019.
21. I-Chen Wu, Ti-Rong Wu, An-Jen Liu, Hung Guei, Tinghan Wei, “On Strength Adjustment for MCTS-Based Programs,” The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19), Hawaii, USA, Jan. 2019. (Acceptance rate: 16.2% among 7,095 papers)
22. Hung Guei, Ting-Han Wei, I-Chen Wu, “Using 2048-like Games as a Pedagogical Tool for Reinforcement Learning,” The 10th International Conference on Computers and Games (CG2018), New Taipei, Taiwan, Jul. 2018.
23. Hung Guei, Tinghan Wei, Jin-Bo Huang, I-Chen Wu, “An Empirical Study on Applying Deep Reinforcement Learning to the Game 2048,” The Workshop Neural Networks in Games in the 9th International Conference on Computers and Games (CG2016), Leiden, the Netherlands, Jun. 2016.

Open Source

1. Ti-Rong Wu, Hung Guei, et al., “MiniZero: An AlphaZero and MuZero Training Framework,” GitHub repository.
2. Yu-Hung Chang*, Hung Guei*, Ti-Rong Wu, “Tutorials for reinforcement learning in games,” GitHub repository. (* contributed equally)
3. Hung Guei, “TDL2048+: The Most Efficient Temporal Difference Learning Framework for 2048,” GitHub repository.
4. Hung Guei, “TDL2048-Demo: Temporal Difference Learning for Game 2048 (Demo),” GitHub repository.