Computational models of learning
BACONcognitive sciencecognitive structuresconceptual knowledgedata compressiondegree of abstractionfunctional relationsheuristic learninginductive learningknowledge engineeringmachine learningmodelling human intelligencequantitative empirical lawsUniversal Puzzle Learner
Conference proceedings and collections of articles (00Bxx) Proceedings, conferences, collections, etc. pertaining to computer science (68-06) Learning and adaptive systems in artificial intelligence (68T05) Problem solving in the context of artificial intelligence (heuristics, search strategies, etc.) (68T20)
The book presents a collection of 5 contributions, investigating various strategies of (human) learning and developing corresponding cognitive models supported by computer simulations. \textit{R. S. Michalski} gives a general overview with emphasis on inductive learning. \textit{P. Langley}, \textit{H. A. Simon} and \textit{G. L. Bradshaw} describe several versions of BACON, a system for finding functional relations between given numerical data and suggesting general principles underlying discovery of quantitative empirical laws. \textit{S. Ohlsson} deals with heuristic learning which occurs when a problem solver has settled on a representation for a task but does not yet know a strategy how to process it. The system UPL (Universal Puzzle Learner) incorporating both task-independent problem solving facilities and task- specific mechanisms (proposers, censors and subgoaling rules) is described and compared with the systems SAGE, LEX, GPS, XAPS3 and GRAPES. \textit{L. Rendell} introduces `conceptual knowledge' measuring the degree of abstraction (information compression), to facilitate automated learning or knowledge acquisition from detailed data (exhibiting little abstraction). The Probability Learning System PLS for induction learning in incremental situations involving uncertainity, is discussed. \textit{J. G. Wolff} applies (Prolog based) grammars to express data compression principles in optimization of various cognitive structures. The papers outline a few of the many ways how the complex problem of modelling human intelligence can be attacked. Although the underlying mathematics is not explicitly developed and the treatment often lacks conciseness of mathematical argumentation (partly due to the fact that many aspects of learning can be at present hardly fully formalized), the book suggests approaches deserving attention for further research even in applied mathematics.
This page was built for publication: Computational models of learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4692793)
