Learning overhypotheses with hierarchical Bayesian models
- 17 April 2007
- journal article
- research article
- Published by Wiley in Developmental Science
- Vol. 10 (3), 307-321
- https://doi.org/10.1111/j.1467-7687.2007.00585.x
Abstract
Inductive learning is impossible without overhypotheses, or constraints on the hypotheses considered by the learner. Some of these overhypotheses must be innate, but we suggest that hierarchical Bayesian models can help to explain how the rest are acquired. To illustrate this claim, we develop models that acquire two kinds of overhypotheses - overhypotheses about feature variability (e.g. the shape bias in word learning) and overhypotheses about the grouping of categories into ontological kinds like objects and substances.Keywords
This publication has 39 references indexed in Scilit:
- Theory-based Bayesian models of inductive learning and reasoningTrends in Cognitive Sciences, 2006
- How children know the relevant properties for generalizing object namesDevelopmental Science, 2002
- Climate and the limits of permafrost: a zonal analysisPermafrost and Periglacial Processes, 2002
- Concept formation in infancyCognitive Development, 1993
- Object Properties and Knowledge in Early Lexical LearningChild Development, 1991
- Object Properties and Knowledge in Early Lexical LearningChild Development, 1991
- Ontological categories guide young children's inductions of word meaning: Object terms and substance termsCognition, 1991
- PRINCIPLES OF OBJECT PERCEPTIONCognitive Science, 1989
- Word Learning in Children: An Examination of Fast MappingChild Development, 1987
- A learning algorithm for boltzmann machinesCognitive Science, 1985