Macro-operators: a weak method for learning
Artificial Intelligence - Lecture notes in computer science 178
Defining operationality for explanation-based learning
Artificial Intelligence
Combining empirical and analytical learning with version spaces
Proceedings of the sixth international workshop on Machine learning
A Study of Explanation-Based Methods for Inductive Learning
Machine Learning
Chunking in Soar: The Anatomy of a General Learning Mechanism
Machine Learning
Explanation-Based Generalization: A Unifying View
Machine Learning
Hi-index | 0.00 |
EBL can learn justified generalizations from only one example when the domain theory is perfect. However, it does not work when the domain theory is imperfect. Imperfectness of the domain theory can be classified into four levels, i.e. incomplete, intractable, inconsistent and non-operational ones. It is necessary to unify EBL and SBL to solve these problems. In this paper, we propose a framework of an augmented EBL to handle plural examples simultaneously. We formalize it on logic program, and introduce a concept of least EBG to extract similarities from plural examples. We discuss on an approach to solve utility problem with the augmented EBL. Utility problem is a problem to learn more efficient description under complete, tractable, consistent but nonoperational domain theories. We define operationality criteria with maximizing usage degree and minimizing backtracking number, and show they increase partial monotonically by generalization. Since this partial monotinicity is not preferable to search operational generalizations, least EBGs are more operational than usual EBGs. We design a simple incremental learner based on least EBGs, and show its usefulness in recursive domain theories. We also discuss on other imperfect theory problems.