|
Reactive search optimization (RSO) defines local-search heuristics based on machine learning, a family of optimization algorithms based on the local search techniques. It refers to a class of heuristics that automatically adjust their working parameters during the optimization phase. RSO methods are at the basis of the Learning and Intelligent Optimization (LION) approach combining machine learning and optimization .〔 〕 ==RSO overview: learning while optimizing== Reactive Search Optimization (RSO), like all local search techniques, is applied to the problem of finding the optimum configuration of a system; such a configuration is usually composed of continuously or discretely varying parameters, while the optimality criterion is a numerical value associated with each configuration. In most cases, an optimization problem can be reduced to finding the (global) minimum of a function whose arguments are the configuration parameters, seen as free variables in the function's domain space. ''Reactive Search Optimization'' advocates the integration of sub-symbolic machine learning techniques into search heuristics for solving complex optimization problems. The word ''reactive'' hints at a ready response to events during the search through an internal ''feedback loop for online self-tuning and dynamic adaptation''. In Reactive Search the history of the search and the knowledge accumulated while moving in the configuration space is used for self-adaptation in an autonomic manner: the algorithm maintains the internal flexibility needed to address different situations during the search, but the adaptation is automated, and executed while the algorithm runs on a single instance and reflects on its past experience. The metaphors for reactive search derive mostly from the individual human experience. Its motto can be "learning on the job". Real-world problems have a rich structure. While many alternative solutions are tested in the exploration of a search space, patterns and regularities appear. The human brain quickly learns and drives future decisions based on previous observations. This is the main inspiration source for inserting online machine learning techniques into the optimization engine of reactive search optimization.〔 Franco Mascia (2007). (Reactive Search and Intelligent Optimization ). 〕 抄文引用元・出典: フリー百科事典『 ウィキペディア（Wikipedia）』 ■ウィキペディアで「Reactive search optimization」の詳細全文を読む スポンサード リンク
|