Minimizing Statistical Bias with Queries.
Abstract
This report describes an exploration criterion that attempts to minimize the error of a learner by minimizing its estimated squared bias. The author describes experiments with locally-weighted regression on two simple kinematics problems, and observed that this "bias-only" approach outperforms the more common "variance-only" exploration approach, even in the presence of noise.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 14, 1995
- Accession Number
- ADA307061
Entities
People
- David A. Cohn
Organizations
- Massachusetts Institute of Technology