Research Interests

Machine Learning, Reinforcement Learning, Transfer Learning, Game AI.

Selected Publications

2014

• Xingguo Chen, Yang Gao, and Shunguo Fan. Temporal Difference Learning with Piecewise Linear Basis. Chinese Journal of Electronics (CJE), 23(1): 49-54, 2014. (SCI) BibTeX
• 陈兴国, 高阳，范顺国，俞亚君. 基于核方法的连续动作Actor-Critic学习. 模式识别与人工智能. (Accepted)

2013

• Xingguo Chen, Yang Gao, and Ruili Wang. Online Selective Kernel-based Temporal Difference Learning. IEEE Transactions on Neural Networks and Learning Systems (IEEE TNNLS), 24(12): 1944-1956, 2013. (SCI) BibTeX
• 陈兴国, 高阳，范顺国，俞亚君. 基于核方法的连续动作Actor-Critic学习. 中国机器学习会议(CCML), 2013.

2012

• Yujing Hu, Yang Gao, Ruili Wang, Zhaonan Sun, and Xingguo Chen. nMetaQ-an n-agent Reinforcement Learning Algorithm based on Meta Equilibrium. In Proc. Adaptive Learning AgentWorkshop in the 12th International Conference on Autonomous Agents and Multi-agent Systems (AAMAS ALA 2012), 87-94, 2012.

2010

• Hao Wang, Yang Gao, and Xingguo Chen. RL-DOT: A Reinforcement Learning NPC Team for Playing Domination Games. IEEE Transactions on Computational Intelligence and AI in Games (IEEE TCIAIG), 2(1): 17-26, 2010. (SCI) BibTeX

2009

• Xingguo Chen, Hao Wang, Weiwei Wang, Yinghuan Shi, and Yang Gao. Apply ant colony optimization to Tetris. In Proc. 11th Annual Conference on Genetic and Evolutionary Computation (GECCO). 1741-1742, Montreal, QC, Canada, 2009. (EI) BibTeX
• 高阳，王巍巍，陈兴国，葛�. 关系强化学习研究：以俄罗斯方块游戏为例. 机器学习及其应用，清华大学出版社, 33-48, 2009. BibTeX
• 高阳，王皓，陈兴国. 强化学习. 中国计算机学会通讯, 5(8): 42-50, 2009. BibTeX
  Yang Gao, Hao Wang, and Xingguo Chen. Reinforcement learning. Communications of the China Computer Federation (CCCF), 5(8): 42-50, 2009. BibTeX

2008

• 王皓，高阳,陈兴国. 强化学习中的迁移：方法和进展. 电子学报，36(12A): 39-43, 2008. BibTeX
  Hao Wang, Yang Gao, and Xingguo Chen, Transfer of reinforcement learning: the state of the art. Chinese Journal of Electronics, vol. 36(12A):39-43, 2008. BibTeX
• Weiwei Wang, Yang Gao and Xingguo Chen. Reinforcement Learning with Markov Logic Networks. In Proceedings of MICAI 2008, LNAI 5317, 230-243, 2008. BibTeX
• Weiwei Wang, Xingguo Chen and Yang Gao. Reinforcement Learning with Markov Logic Networks. EWRL'08 Presentation, 2008.
• 王巍巍,陈兴国,高阳.一种结合Tile Coding的平均奖赏强化学习算法. 模式识别与人工智能, 21(4): 446-452, 2008.
  Weiwei Wang, Xingguo Chen and Yang Gao. Approximation methods based on average reward learning with Tile Coding. Pattern Recognition and Artificial Intelligence, 21(4): 446-452,2008.
• 陈兴国. Tetris问题中的几种机器学习方法比较. 第三届机器学习及其应用学生研讨会,2008. (Poster)

Awards & Honors

• 南瑞奖学金, 南京大学, 2013.
  Nanrui Scholarship, Nanjing University, 2013.
• 最佳论文提名奖, 第十四届中国机器学习会议, 昆明, Aug. 2013. (基于核方法的连续动作Actor-Critic学习)
  Best Paper Honorable Mention Award, the 14th China Conference on Machine Learning, Kunming, Aug. 2013.
• 优秀研究生, 南京大学, 2011.
  Excellent Postgraduate Scholarship, Nanjing University, 2011.
• 南瑞奖学金, 南京大学, 2010.
  Nanrui Scholarship, Nanjing University, 2010.
• 真知味奖学金, 南京大学, 2009.
  Zhenzhiwei Scholarship, Nanjing University, 2009.
• Second place on Tetris event, Reinforcement Learning Competition, 2009. (NJU RL Team)
• 人民奖学金三等奖, 南京大学，2006.
  Third Prize of the People's Scholarship, Nanjing University, 2006.
• 人民奖学金三等奖, 南京大学，2005.
  Third Prize of the People's Scholarship, Nanjing University, 2005.
• 江苏省赛区一等奖, 全国大学生数学建模竞赛，2004.
  First prize in Jiangsu Province, China Undergraduate Mathematical Contest in Modeling, 2004
• 人民奖学金三等奖, 南京大学，2004.
  Third Prize of the People's Scholarship, Nanjing University, 2004.

AI Tetris

Tetris is a falling-blocks puzzle video game originally designed and programmed by Alexey Pajitnov in 1985. In Tetris, a pseudorandom sequence of tetrominoes (sometimes called "tetrads" in older versions) - shapes composed of four square blocks each - fall down the playing field. The object of the game is to manipulate these tetrominoes by moving each one sideways and rotating it by 90 degree units, with the aim of creating a horizontal line of blocks without gaps. When such a line is created, it disappears, and the blocks above (if any) fall. The game ends when the stack of tetrominoes reaches the top of the playing field and no new tetrominoes are able to enter. Despite its simple rules, playing tetris well requires a complex strategy and lots of experience (For more information, see Tetris).

Below is a demon of Tetris in Java Applet (to run, you have to install jdk), where agent plays Tetris game automatically. The policy of the agent is based on a linear approximated after-state value function, where the weight has been learned by machine learning algorithms. You can change the weight to test.

Professional Activities

Friends

Dong Kai, Ge Shen, Guo Qiaojin, Huang Shujian, Ji Yangsheng, Shi Liangdong , Shi Yinghuan , Wang Hao, Wang Liang, Wang Weiwei, Xi Ning, Xu Tianyin, Zhan Andong