141 research outputs found
A Study on Comprehensibility of Concept Descriptions
No full textA transformation of descriptions of concepts without a change of meaning makes sense under a heterogeneous space of concepts, like Explanation-Based Learning (EBL) under the operational criterion. Under finite awareness the heterogeneousness is necessary, because agents have to select certain finite sets of concepts under criterion. Comprehensibility seems to come from the finite awareness, and it should be realized by such a transformation of descriptions. In this paper, we discuss comprehensibility of descriptions using a formalism of first order logic with finite awareness, and a couple of characterisations. Methods to transform descriptions in a sense of goodness discussed in the paper. The address is temporary. After March 1996, author's address will be: Department of Intelligence and Computer Science, Nagoya Institute of Technology, Gokiso-cho, Showa-ku, Nagoya 466, Japan. E-mail: [email protected] 1 INTRODUCTION We sometimes find it difficult to comprehend descriptio..
Learning without Model-theoretic Improvements
No full text1 Introduction In this paper, we investigate a different and new interpretation of semantics of expressed concepts in order to learn concepts. In the semantic aspects, learning of concepts has been treated as Model-Theoretic improvements. Model-Theoretically speaking, knowledge is a thing to restrict models of concepts. To learn concepts is regarded to specify models of the concepts. This aspect of view for learning yields an attitude that we don't look upon changes of knowledge without changes of models as learning. Nevertheless, there exist some important and interesting changes of knowledge without changes of models. The most popular one is the Explanationbased Learning (EBL)[1], which makes changes to a theory that improves processing speed. But, it doesn't make any contribution to the models of concepts. We can solve this contradiction and develop a new paradigm of learning by some modification of the Model-Theoretic view. 2 Generalizing Model Theory 2.1 Notation The EBL makes ..
Multi-relational Pattern Mining Based-on Combination of Properties with Preserving Their Structure in Examples
Get PDFjournal articl
Solving selection problems using preference relation based on Bayesian learning
No full textAbstract. This paper defines a selection problem which selects an appropriate object from a set that is specified by parameters. We discuss inductive learning of selection problems and give a method combining inductive logic programming (ILP) and Bayesian learning. It induces a binary relation comparing likelihood of objects being selected. Our methods estimate probability of each choice by evaluating variance of an induced relation from an ideal binary relation. Bayesian learning combines a prior probability of objects and the estimated probability. By making several assumptions on probability estimation, we give several methods. The methods are applied to Part-of-Speech tagging.
- …
