»ã±¨Ö÷Ì⣺¼ÓÇ¿½ø½¨
Reinforcement Learning: An Introduction
±¨ ¸æ ÈË£ºBernard Manderick
±ÈÀûʱ²¼Â³Èü¶û×ÔÓÉ´óѧ ÈËΪÖÇÄܳ¢ÊÔÊÒ ½ÌÊÚ
»ã±¨¹¦·ò£º2013Äê 11Ô 15ÈÕ£¨ÖÜÎ壩ÏÂÎç2:00
»ã±¨µØÖ·£ºÐ±¦GGУ±¾²¿¶«ÇøÍÆËã»ú´óÂ¥1104ÊÒ
Ñû Çë ÈË£ºÁõÔà ²©Ê¿
»ã±¨¼ò½é£º
In this talk we explore a computational approach to learning from interaction. That is, we adopt the perspective of an artificial intelligence researcher. We explore designs for machines that are effective in solving learning problems of scientific, evaluating the designs through mathematical analysis or computational experiments. The approach we explore, called reinforcement learning, is focused on goal-directed learning from interaction.
»ã±¨È˼ò½é£º
Since 1994, Bernard Manderick is professor in the Artificial Intelligence Lab at the Department of Computer Science of the Vrije Universiteit Brussel. Currently, the AI Lab consists of 3 professors, 3 senior researchers, 8 postdocs, and 15 PhD students. For more information, cf. the homepage of the AI Lab http://ai.vub.ac.be for more information. He is (co-)author of over 130 papers covering several machine learning techniques (genetic algorithms, support vector machines, reinforcement learning, Bayesian networks, … ) and several of its applications (bioinformatics, text mining, evolvable hardware, music classification, …). And, he is (co)-supervisor of 11 PhD-theses and currently, he (co)-supervises 8 PhD students. Finally, he was/is coordinator of over 25 research projects and is involved in several international research co-operations.
For his PhD-thesis Selectionism as a Basis for Categorization and Adaptation, Bernard Manderick received the IBM–prize for Informatics awarded by the Fund for Scientific Research (FWO). He also received a Science and Technology Agency Fellowship from the European Commission (DG XII) sponsored by the Japanese Government and an ERCIM (European Research Center for Informatics and Mathematics) Fellowship.
