It has been suggested that AI investigations of mechanical learning undermine sweeping anti-inductivist views in the theory of knowledge and the philosophy of science. Contrary to this view, we argue that no trace of epistemically jus- tified induction is to be found within a rather representative class of learning agents and we outline an alternative deductive account of these learning procedures. Finally, the opportunity of developing an induction-free logical analysis of reasoning processes involved in learning machines is emphasized by a broad reflection on some families of non-monotonic, albeit deductive, consequence relations. Keywords: Induction,machine learning, behavior-based robotics, non-monotonic consequence relations, artificial intelligence.
An inductionless and default-based analysis of machine learning procedures / E. Datteri, H. Hosni, G. Tamburrini (STUDIES IN LOGIC). - In: Model Based Reasoning in Science and Engineering / [a cura di] L. Magnani. - [s.l] : College Publications London, 2006. - ISBN 1904987230. - pp. 379-400 (( convegno Model-Based Reasoning in Science and Engineering: Abduction, Visualization, Simulation tenutosi a Pavia nel 2004.
An inductionless and default-based analysis of machine learning procedures
H. Hosni;
2006
Abstract
It has been suggested that AI investigations of mechanical learning undermine sweeping anti-inductivist views in the theory of knowledge and the philosophy of science. Contrary to this view, we argue that no trace of epistemically jus- tified induction is to be found within a rather representative class of learning agents and we outline an alternative deductive account of these learning procedures. Finally, the opportunity of developing an induction-free logical analysis of reasoning processes involved in learning machines is emphasized by a broad reflection on some families of non-monotonic, albeit deductive, consequence relations. Keywords: Induction,machine learning, behavior-based robotics, non-monotonic consequence relations, artificial intelligence.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.