Reinforcement Distribution in Continuous State Action Space Fuzzy Q–Learning: A Novel Approach Export

@incollection{citeulike:1723216, abstract = {Fuzzy Q–learning extends the Q–learning algorithm to work in presence of continuous state and action spaces. A Takagi–Sugeno Fuzzy Inference System (FIS) is used to infer the continuous executed action and its action–value, by means of cooperation of several rules. Different kinds of evolution of the parameters of the FIS are possible, depending on different strategies of distribution of the reinforcement signal. In this paper, we compare two strategies: the classical one, focusing on rewarding the rules that have proposed the actions composed to produce the actual action, and a new one we are introducing, where reward goes to the rules proposing actions closest the ones actually executed.}, author = {Bonarini, Andrea and Montrone, Francesco and Restelli, Marcello}, citeulike-article-id = {1723216}, citeulike-linkout-0 = {http://dx.doi.org/10.1007/11676935_5}, comment = {L'articolo propone un nuovo approccio al Fuzzy-qlearning utilizzando una particolare fuzzificazione della funzione di valore delle coppie stato azione. Lo scopo rimane cmq quello di poter trattare l'apprendimento per rinforzo anche in casi con spazi degli stati e delle azioni continui, utilizzando i metodi fuzzy per costruire un approssimatore di funzione.}, doi = {10.1007/11676935_5}, journal = {Fuzzy Logic and Applications}, keywords = {continuous\_time, fuzzy, learning, reinforcement}, pages = {40--45}, posted-at = {2007-10-03 09:16:30}, priority = {0}, title = {Reinforcement Distribution in Continuous State Action Space Fuzzy Q–Learning: A Novel Approach}, url = {http://dx.doi.org/10.1007/11676935_5}, year = {2006} }

TY - CHAP ID - citeulike:1723216 L3 - citeulike-article-id:1723216 N2 - Fuzzy Q–learning extends the Q–learning algorithm to work in presence of continuous state and action spaces. A Takagi–Sugeno Fuzzy Inference System (FIS) is used to infer the continuous executed action and its action–value, by means of cooperation of several rules. Different kinds of evolution of the parameters of the FIS are possible, depending on different strategies of distribution of the reinforcement signal. In this paper, we compare two strategies: the classical one, focusing on rewarding the rules that have proposed the actions composed to produce the actual action, and a new one we are introducing, where reward goes to the rules proposing actions closest the ones actually executed. JF - Fuzzy Logic and Applications EP - 45 TI - Reinforcement Distribution in Continuous State Action Space Fuzzy Q–Learning: A Novel Approach SP - 40 KW - continuous_time KW - fuzzy KW - learning KW - reinforcement N1 - L'articolo propone un nuovo approccio al Fuzzy-qlearning utilizzando una particolare fuzzificazione della funzione di valore delle coppie stato azione. Lo scopo rimane cmq quello di poter trattare l'apprendimento per rinforzo anche in casi con spazi degli stati e delle azioni continui, utilizzando i metodi fuzzy per costruire un approssimatore di funzione. AU - Bonarini, Andrea AU - Montrone, Francesco AU - Restelli, Marcello PY - 2006/// UR - http://dx.doi.org/10.1007/11676935_5 ER -