Book Chapter10.1016/B978-1-55860-036-2.50111-9
Knowledge-based feature generation
James P. Callan
- 01 Dec 1989
- pp 441-443
8
TL;DR: A method of automatically enriching an instance description language with new, problem-specific terms is described, independent of any particular learning algorithm or concept description language, guided by whatever domain-specific information is available, such as a search problem specification.
read more
Abstract: The instance description language of machine learning algorithms is usually crafted by humans. This paper describes a method of automatically enriching an instance description language with new, problem-specific terms. The method is independent of any particular learning algorithm or concept description language. Instead, it is guided by whatever domain-specific information is available, such as a search problem specification. An example shows how two useful new terms can be created for the n-queens problem.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Learning hard concepts through constructive induction: framework and rationale
Larry A. Rendell,Raj Seshu +1 more
- 02 Jan 1991
TL;DR: This work argues for a specific approach to constructive induction that reduces variation by incorporating various kinds of domain knowledge, i.e., transformations that group together non‐contiguous portions of feature space having similar class‐membership values.
147
•Journal Article
Harnessing the Expertise of 70,000 Human Editors: Knowledge-Based Feature Generation for Text Categorization
TL;DR: This work presents a new framework for automatic acquisition of world knowledge and methods for incorporating it into the text categorization process, which enhances machine learning algorithms with features generated from domain-specific and common-sense knowledge.
•Proceedings Article
Constructive induction on domain information
James P. Callan,Paul E. Utgoff +1 more
- 14 Jul 1991
TL;DR: It is shown that information about the problem-solving task can be used to create terms that are suitable for learning search control knowledge, and the resulting terms describe theproblem-solver's progress in achieving its goals.
12
Knowledge-based feature generation for inductive learning
James Patrick Callan
- 01 Jan 1993
TL;DR: This dissertation develops knowledge-based feature generation, a stronger, but more restricted, method of constructive induction than was available previously, and shows knowledge- based feature generation to be a general method of creating useful new features for one class of learning problems.
10
Feature Discovery for Inductive Concept Learning
Tom E. Fawcett
- 01 Feb 1990
TL;DR: Zenith is a discovery system that performs constructive induction that is able to generate and extend new features for concept learning using agenda-based heuris- tic search and its ability to extend its knowledge base by creating new domain classes is distinguished.
8
References
A theory and methodology of inductive learning
TL;DR: The authors view inductive learning as a heuristic search through a space of symbolic descriptions, generated by an application of various inference rules to the initial observational statements, including generalization rules, which perform generalizing transformations on descriptions, and conventional truth-preserving deductive rules.
1.3K
A problem similarity approach to devising heuristics: first results
John Gaschnig
- 20 Aug 1979
TL;DR: Evidence of a role for exploiting certain similarities among problems to transfer a heuristic from one problem to another, from an "easier" problem to a "harder" one is presented.
68
Evaluating alternative instance representations
Sharad Saxena
- 01 Dec 1989
TL;DR: An algorithm for estimating the complexity of the function from a set of examples is proposed and was able to correctly identify the better of the two given representations for the two-or-more-clumps problem.
6
Related Papers (5)
Peter Norvig
- 01 Jan 1986
Carl G. De Marcken,Robert C. Berwick +1 more
- 01 Jan 1996
David D. McDonald
- 29 Jun 1981
Jacob Andreas,Dan Klein,Sergey Levine +2 more
- 01 Jun 2018