Learning One Subprocedure per Lesson.
Abstract
Sierra is a program that learns procedures incrementally from examples. It learns by completing explanations. Whenever the current procedure is inadequate to explain (parse) the current example (a string of actions), Sierra formulates a new subprocedure whose instantiation completes the explanation. This technique is one solution to the problem of acquiring the domain theory required by learning-by explanation systems. The key to its success lies in supplying a small amount of extra information with the examples. Instead of giving the learner an set of examples, under which conditions correct learning is provably impossible, it is given a sequence of lessons, where a lesson is a set of examples, possibly a singleton set, that is guaranteed to introduce only one subprocedure. This permits unbaised learning, i.e., learning without a priori, heuristic preferences concerning the outcome. Keywords: Learning, skill acquisition, machine learning, induction, grammatical inference, felicity conditions, explanation-based learning.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 13, 1985
- Accession Number
- ADA169360
Entities
People
- Kurt A. Vanlehn
Organizations
- PARC