Abstract :
Machine-based learning will eventually be applied to solve real-world problems. In this work, an associative architecture teams up with hybrid AI algorithms to solve a letter prediction problem with promising results. This article describes an investigation and simulation of a massively parallel learning classifier system (LCS) that was developed from a specialized associative architecture joined with hybrid AI algorithms. The LCS algorithms were specifically invented to computationally match a massively parallel computer architecture, which was a special-purpose design to support the inferencing and learning components of the LCS. The LCS´s computationally intensive functions include rule matching, parent selection, replacement selection and, to a lesser degree, data structure manipulation.<>
Keywords :
associative processing; character recognition; genetic algorithms; learning (artificial intelligence); parallel algorithms; parallel architectures; pattern classification; associative architecture; computational matching; computationally intensive functions; data structure manipulation; genetic algorithm-based machine learning; hybrid AI algorithms; inferencing; letter prediction; massively parallel computer architecture; massively parallel learning classifier system; parent selection; real-world problem solving; replacement selection; rule matching; simulation; special-purpose design; Artificial intelligence; Computational modeling; Computer architecture; Concurrent computing; Genetics; Inference algorithms; Learning systems; Machine learning; Machine learning algorithms; Production systems;
