摘要：AbstractIn any online adaptation scheme, two important phenomena should be taken into consideration; parameter shadowing and parameter interference. To alleviate these problems, in this paper a convex hull, sliding window based online adaptation method for fixed-structure Neural Networks is proposed. The method is capable of dealing with the two phenomena, presenting better results than known alternatives. An analysis of the real-time run time and memory consumption of the algorithm demonstrates that it can be used for real-time applications.
关键词：KeywordsConvex HullMulti Objective Genetic AlgorithmOnline Adaptation ProcessRadial Basis Function Neural NetworksTime Series Models
