在复杂系统研究领域，一直存在着对经验、直觉等认知建模的需求，由于缺乏对认知进行有效建模和仿真的手段，这一问题已成为对复杂系统整体涌现性、混沌、不确定性等特性深入理解的主要瓶颈。分析了“阿尔法狗”在认知智能上的突破，阐述了认知仿真的基本内涵，探讨了经验直觉捕捉对复杂系统建模的重要意义，提出了认知仿真方法依然需要深入思考的问题。
In the complex system researches, the cognitive modeling of experience and intuition is always desirable. Due to the lack of effective modeling and simulation methods for the cognition, this problem becomes a major bottleneck restricting the overall emergence, the chaos and the uncertainty of the complex systems. This paper analyzes the breakthrough of AlphaGo in the cognitive intelligence, as well as the basic connotations of the cognitive simulation, and points out the importance of the empirical and intuition capture for modeling complex systems and. the issues that the cognitive simulation methods should consider.
[1] Silver D, Huang A, Maddison C J, et al. Mastering the game of Go with deep neural networks and tree search[J]. Nature, 2016, 529(7587):484-489.
[2] Borow H. Information retrieval:Ⅱ. Simple and complex systems.[J]. Journal of Counseling Psychology, 1963, 10(1):88-93.
[3] Grabowski F, Strzalka D. Simple, complicated and complex sys-tems-The brief introduction[C]//Human System Interactions, 2008 Conference on IEEE, 2008:570-573.
[4] Feistel R, Ebeling W. Evolution of complex systems:Self orga-nization, entropy and development[M]. Holland:Kluwer Aca-demic Publisher, 1989.
[5] 约翰·H·霍兰. 隐秩序:适应性造就复杂性[M]. 周晓牧, 等, 译. 上海:上海科技教育出版社, 2011. Horan J H. Hidden order:Adaptability create complexity[M]. Zhou Xiaomu, et al., trans. Shanghai:Shanghai Science and Technology Education Press, 2011
[6] Holland J H. Complex adaptive systems[J]. Daedalus, 1992, 121(1):17-30.
[7] Lansing J S. Complex adaptive systems[J]. Annual Review of Anthropology, 2003, 32(4):183-204.
[8] 中国科协学会学术部. 复杂系统建模仿真中的困惑和思考[M]. 北京:中国科学技术出版社, 2012. Department of Societies and Academic, China Association for Science and Technology. Bewilderment and thinking for com-plex system modeling and simulation[M]. Beijing:China Sci-ence and Technology Press, 2012.
[9] 中国科协学会学术部. 大数据时代对建模仿真的挑战与思考[M]. 北京:中国科学技术出版社, 2014. Department of Societies and Academic, China Association for Science and Technology. Challenges and reflections on model-ing and simulation in the era of large data[M]. Beijing:China Science and Technology Press, 2014
[10] Helleboogh A, Vizzari G, Uhrmacher A, et al. Modeling dy-namic environments in multi-agent simulation[J]. Autono-mous Agents and Multi-agent Systems, 2007, 14(1):87-116.
[11] Messina F, Pappalardo G, Santoro C. ComplexSim:An smpaware complex network simulation framework[C]//Sixth inter-national conference on complex, intelligent, and software in-tensive systems. IEEE Computer Society, 2012:861-866.
[12] Ranjan R. Modeling and simulation in performance optimiza-tion of big data processing frameworks[J]. IEEE Cloud Com-puting, 2015, 1(4):14-19.
[13] 王飞跃. 平行系统方法与复杂系统的管理和控制[J]. 控制与决策, 2004, 19(5):485-489. Wang Feiyue. Parallel system method and complex system management and control[J]. Control and Decision, 2004, 19(5):485-489.
[14] Mnih V, Kavukcuoglu K, Silver D, et al. Playing atari with deep reinforcement learning[J]. arXiv preprint arXiv:1312.5602, 2013.
[15] Moravcík M, Schmid M, Burch N, et al. DeepStack:Expertlevel artificial intelligence in heads-up no-limit poker[J]. Sci-ence, 2017, 356(6337):508-513.
[16] Brown N, Sandholm T. The Superhuman AI for no-limit poker[C]//Twenty-Sixth International Joint Conference on Artificial Intelligence, 2017:5226-5228.
[17] Ontanon S, Synnaeve G, Uriarte A, et al. A Survey of realtime strategy game AI research and competition in starcraft[J]. IEEE Transactions on Computational Intelligence & Ai in Games, 2013, 5(4):293-311.
[18] Krizhevsky A, Sutskever I, Hinton G E. ImageNet classifica-tion with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems. Curran Associates Inc. 2012:1097-1105.
[19] Gers F A, Schmidhuber J, Cummins F. Learning to forget:con-tinual prediction with LSTM[J]. Neural Computation, 2000, 12(10):2451.
[20] Hinton G E. Deep belief networks[J]. Scholarpedia, 2009, 4(6):5947.
[21] Weber T, Racanière S, Reichert D P, et al. Imagination-aug-mented agents for deep reinforcement learning[J]. arXiv:1707.06203v2[cs.LG] for this version
[22] Pascanu R, Li Y, Vinyals O, et al. Learning model-based planning from scratch[J]. arXiv preprint arXiv:1707.06170, 2017.
[23] Paul Voosen, How AI detectives are cracking open the black box of deep learning[J]. Science 2017, doi:10.1126/science. aan7059.
