A Self-Organized, Distributed, and Adaptive Rule-Based Induction System

Abstract

Learning classifier systems (LCSs) are rule-based inductive learning systems that have been widely used in the field of supervised and reinforcement learning over the last few years. This paper employs sUpervised Classifier System (UCS), a supervised learning classifier system, that was introduced in 2003 for classification tasks in data mining. We present an adaptive framework of UCS on top of a self-organized map (SOM) neural network. The overall classification problem is decomposed adaptively and in real time by the SOM into subproblems, each of which is handled by a separate UCS. The framework is also tested with replacing UCS by a feedforward artificial neural network (ANN). Experiments on several synthetic and real data sets, including a very large real data set, show that the accuracy of classifications in the proposed distributed environment is as good or better than in the nondistributed environment, and execution is faster. In general, each UCS attached to a cell in the SOM has a much smaller population size than a single UCS working on the overall problem; since each data instance is exposed to a smaller population size than in the single population approach, the throughput of the overall system increases. The experiments show that the proposed framework can decompose a problem adaptively into subproblems, maintaining or improving accuracy and increasing speed.

Market segmentation and ideal point identification for new product design using fuzzy data compression and fuzzy clustering methods

“…As the dimensions of   i w is only against the two principle components, f 1 and f 2 , it is necessary to map the ideal point,   i w , from two dimensions into d dimensions with respect to the d desired customer requirements. Equations (10) to (12) can be used to map the ideal point,…”

“…It assesses the rank-order or overall value for alternatives with different profiles of attribute levels, and then uses the holistic judgment information to estimate discrete levels of single-attribute value functions by regressions, hierarchical Bayes models, or linear programming [8]. Self-organized feature map is widely used for dimension reduction and clustering, concurrently for various applications of which the data is in multi-dimensions [9][10][11]. However, this approach has only been used on processing real data, while processing data in fuzzy numbers has not been addressed.…”

ReAd: reactive-adaptive methodology to enable evolving intelligent agents for virtual environments

“…In Rojanavasu et al (2009) the authors use adaptive decomposition in real time to create a fast and accurate supervised learning classifier system; in Kolodyazhniy and Bodyanskiy (2010) the authors introduce a cascaded multi-resolution splinebased fuzzy neural network and a fast constructive training algorithm for real-time time-series prediction. Taking all these advances into consideration, ANN based systems are still more computationally expensive than a FSM or BDI when executing multiple individual processes in parallel; for process control operations, robot control or a limited number of NPCs, ANNs can be an excellent choice, but when the number of individual instances (agents) running in parallel in the same simulated world increases (to dozens or more), then the small computational cost differences become more apparent and eventually significant.…”

An adaptive knowledge evolution strategy for finding near-optimal solutions of specific problems

“…Moreover, learning classifier systems that were first described by Prof. John Holland consist of computing rules which are composed of binary, real-valued, neural network, and other representations. Recently, researchers have proposed an enhancement using self-organized map (SOM) [33] and neural network [34] on supervised learning classifier systems (UCS). UCS is also used for the classification problem in data mining [35].…”