Machine Learning
Abstract
Our goal is to design and implement an intelligent planner that improves its own performance as it solves problems. This system, called ULS, is based upon explanation-based learning techniques, but addresses some of the weaknesses in that technology that become particularly apparent as it is applied to realistic problems. Our research has been specifically directed toward the above-mentioned utility problem. Our approach to addressing the utility of learning is to approximate the results of explanation-based learning (Cha89, Zwe88). This year, our efforts were focused on improving the approximation techniques, defining new simplified domains, and conducting experiments in various domains.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1990
- Accession Number
- ADA223732
Entities
People
- Melissa P. Chase
Organizations
- MITRE Corporation