Genetic Plans and the Probabilistic Learning System: Synthesis and Results
Larry A. Rendell · 1985
This paper describes new conceptual and experimental results using the probabilistic learning system PLSe, PLS2 is designed ror any task in which overall performance can be measured, and in which choice or task objects or operators influences perrormance, The system can manage incremental learning and noisy domains. PLS2 learns in two ways. Its lower "perceptual" layer clusters data in to economical cells or regions in augmented reature space. The upper "genetic" level or PLS2 selects successful regions (compressed genes) from multiple, parallel cases. Intermediate between performance data and task control structures, regions promote efficient and effective learning. Novel aspects or PLS2 include compressed genotypes, credit localization and "population performance", Incipient principles or efficiency and effectiveness are suggested. Analysis of the system is confirmed by experiments demonstrating stability, efficiency, and effectiveness. 1