Parallel exploration via negatively correlated search
作者机构：Guangdong Provincial Key Laboratory of Brain-inspired Intelligent ComputationDepartment of Computer Science and EngineeringSouthern University of Science and TechnologyShenzhen 518055China
出 版 物：《Frontiers of Computer Science》 (中国计算机科学前沿（英文版）)
年 卷 期：2021年第15卷第5期
页 面：123-135页
核心收录：
学科分类：0810[工学-信息与通信工程] 0808[工学-电气工程] 08[工学] 0701[理学-数学] 081201[工学-计算机系统结构] 0812[工学-计算机科学与技术（可授工学、理学学位）]
基 金：the Natural Science Foundation of China(Grant Nos.61806090 and 61672478) Guangdong Provincial Key Laboratory(2020B121201001) the Program for Guangdong Introducing Innovative and Entrepreneurial Teams(2017ZT07X386) the Science and Technology Commission of Shanghai Municipality(19511120600) Shenzhen Science and Technology Program(KQTD2016112514355531)
主 题：evolutionary computation reinforcement learning exploration
摘 要：Effective exploration is key to a successful search *** recently proposed negatively correlated search(NCS)tries to achieve this by coordinated parallel exploration,where a set of search processes are driven to be negatively correlated so that different promising areas of the search space can be visited *** successful applications of NCS,the negatively correlated search behaviors were mostly devised by intuition,while deeper(e.g.,mathematical)understanding is *** this paper,a more principled NCS,namely NCNES,is presented,showing that the parallel exploration is equivalent to a process of seeking probabilistic models that both lead to solutions of high quality and are distant from previous obtained probabilistic *** learning,for which exploration is of particular importance,are considered for empirical *** proposed NCNES is applied to directly train a deep convolution network with 1.7 million connection weights for playing Atari *** results show that the significant advantages of NCNES,especially on games with uncertain and delayed rewards,can be highly owed to the effective parallel exploration ability.
维普期刊数据库