Title: Improvement of Perception through Task Executions
Authors: Ryohei Orihara, Tomoko Murakami, Takehiko Yokota, and Naomichi Sueda
Series: Linköping Electronic Articles in Computer and Information Science
ISSN 1401-9841
Issue: Vol. 5 (2000), No. 033
URL: http://www.ep.liu.se/ea/cis/2000/033/

Abstract: Good perception is indispensable to perform problem solving well. This is a common issue in cognitive psychology, but has been seldom studied in artificial intelligence. We formalize a perception scheme for an intelligent software using vector spaces. We also construct an experimental system for text documents and still images. Using the system for the texts, we conduct experiments in a context of information retrieval. The experiments using FAQ (Frequently Asked Questions) documents prove the method to be outperforming the conventional method. As for the system for the images, we extract common features out of a certain set of images in terms of Kansei engineering. The features explain characteristics of the images in intuitive manner. We aim to construct a decision support system, which cna analyze multimedia data from various viewpoints; however we will need to extend the method in various ways to achieve that.

Original publication
Postscript part I -- Checksum
Checksum (old) Information about recalculation of checksum
Postscript part II -- Checksum II
Checksum II (old) Information about recalculation of checksum