Selected work

Representative Papers

Temporal-Difference Learning for Dragonchess

Oct 01, 2026

This study addresses the challenge of effectively updating evaluation heuristics within the complex state space of 3D Dragon Chess. Leveraging a high-performance engine reconstructed in C++, we systematically compare two adaptive learning strategies: evolutionary transfer learning and TD(λ). Through extensive match statistics and significance testing, experimental results demonstrate that both methods significantly outperform baseline agents, with no statistically significant performance difference observed between them. This research confirms the efficacy of adaptive strategies in novel game domains characterized by structural complexity, providing reliable methodological support for AI-driven solutions to high-dimensional games with incomplete information.

0 citationsRead paper

Niching Agents in The Core

Sep 10, 2026

研究通过在特定子环境中划分代理，利用The Core算法的局部交互、锦标赛选择、交叉和变异方法，进化出更成功的自主控制代理。

0 citationsRead paper

Decentralized Evolution of Hexapod Gaits with Independent Leg Controllers

Sep 10, 2026

本文通过独立进化每条腿的步态，采用去中心化的进化算法解决六足机器人行走问题，优化了个体腿部控制器以实现高效协调的运动。

0 citationsRead paper

DSLE: A Learning Environment for Dark Souls Boss Encounters

Aug 10, 2026

This work addresses the challenge of learning effective combat strategies in realistic action games with high-dimensional visual inputs and sparse rewards. The authors propose DSLE, the first containerized platform that systematically converts all 22 boss fights from *Dark Souls: Remastered* into Gymnasium-compatible environments, and introduce DSLE-5—a benchmark suite comprising five representative scenarios. The platform supports training via pixel inputs using mainstream methods such as PPO, DQN, evolutionary algorithms, and expert systems. Experimental results show that only expert systems and evolutionary algorithms achieve non-trivial success against the tutorial boss (peak win rate: 63%), while standard deep reinforcement learning approaches fail to learn effectively. Even with a max-level character, existing methods struggle to defeat most bosses, highlighting the substantial difficulty of policy learning in authentic game environments.

0 citationsRead paper

Evolutionary Transfer Learning for Dragonchess

Mar 16, 2026

This study addresses the challenge of effectively transferring heuristic knowledge from the chess engine Stockfish to Dragonchess, a three-dimensional board game with substantially different rules and structure. To this end, we introduce Dragonchess as a novel benchmark for AI transfer learning and propose an approach that employs the CMA-ES evolutionary strategy to optimize the heuristic evaluation function after transfer. We also develop an open-source Python engine to facilitate experimentation. In a 50-round Swiss-system tournament, the evolutionarily tuned AI agent significantly outperformed a baseline policy using direct heuristic transfer, demonstrating the feasibility of adapting domain-specific heuristics across dissimilar game environments. This work establishes a new paradigm for transfer learning in complex board games.

0 citationsRead paper

Recent publications

Latest Papers

Temporal-Difference Learning for Dragonchess

Oct 01, 2026

This study addresses the challenge of effectively updating evaluation heuristics within the complex state space of 3D Dragon Chess. Leveraging a high-performance engine reconstructed in C++, we systematically compare two adaptive learning strategies: evolutionary transfer learning and TD(λ). Through extensive match statistics and significance testing, experimental results demonstrate that both methods significantly outperform baseline agents, with no statistically significant performance difference observed between them. This research confirms the efficacy of adaptive strategies in novel game domains characterized by structural complexity, providing reliable methodological support for AI-driven solutions to high-dimensional games with incomplete information.

0 citationsRead paper

Niching Agents in The Core

Sep 10, 2026

研究通过在特定子环境中划分代理，利用The Core算法的局部交互、锦标赛选择、交叉和变异方法，进化出更成功的自主控制代理。

0 citationsRead paper

Decentralized Evolution of Hexapod Gaits with Independent Leg Controllers

Sep 10, 2026

本文通过独立进化每条腿的步态，采用去中心化的进化算法解决六足机器人行走问题，优化了个体腿部控制器以实现高效协调的运动。

0 citationsRead paper

DSLE: A Learning Environment for Dark Souls Boss Encounters

Aug 10, 2026

This work addresses the challenge of learning effective combat strategies in realistic action games with high-dimensional visual inputs and sparse rewards. The authors propose DSLE, the first containerized platform that systematically converts all 22 boss fights from *Dark Souls: Remastered* into Gymnasium-compatible environments, and introduce DSLE-5—a benchmark suite comprising five representative scenarios. The platform supports training via pixel inputs using mainstream methods such as PPO, DQN, evolutionary algorithms, and expert systems. Experimental results show that only expert systems and evolutionary algorithms achieve non-trivial success against the tutorial boss (peak win rate: 63%), while standard deep reinforcement learning approaches fail to learn effectively. Even with a max-level character, existing methods struggle to defeat most bosses, highlighting the substantial difficulty of policy learning in authentic game environments.

0 citationsRead paper

Evolutionary Transfer Learning for Dragonchess

Mar 16, 2026

This study addresses the challenge of effectively transferring heuristic knowledge from the chess engine Stockfish to Dragonchess, a three-dimensional board game with substantially different rules and structure. To this end, we introduce Dragonchess as a novel benchmark for AI transfer learning and propose an approach that employs the CMA-ES evolutionary strategy to optimize the heuristic evaluation function after transfer. We also develop an open-source Python engine to facilitate experimentation. In a 50-round Swiss-system tournament, the evolutionarily tuned AI agent significantly outperformed a baseline policy using direct heuristic transfer, demonstrating the feasibility of adapting domain-specific heuristics across dissimilar game environments. This work establishes a new paradigm for transfer learning in complex board games.

0 citationsRead paper