Looking for a new AI paradigm: Evolving connectionist and fuzzy connectionist systems—Theory and applications for adaptive, on-line intelligent systems

摘要: The paper introduces one paradigm of neuro-fuzzy techniquesand an approach to building on-line, adaptive intelligent systems. This is called evolving conncctionist systems (ECOS). ECOS evolve through incremental, online learning, both supervised and unsupervised. They can accommodate new input data, including features, classes, etc. framework presented illustrated on a particular type neural networks ~ fuzzy networks. are three six orders magnitude faster than the multilayer perceptrons, or networks, trained with backpropagation algorithm, genetic programming technique. belong generation several real world problems for adaptive, on-line classitication, prediction, decision making control: phoneme-based speech recognition; moving person identification; wastewater time-series prediction control; agents; financial time series control. ’I‘ he principles recurrent reinforcement learning outlined.