Abstract
Recently, many studies about recognition or understanding have managed the multimedia, such as sound and image. In these study, it is very difficult to relate the different kinds of information that have the same meaning, because of each media has appropriate description method in own domain. This paper describes the knowledge representation method to manage the different kinds of information, and that enables acquiring the knowledge by learning. This system can handle two different kinds of information: image and sound. In this system, these informations are related using the rule that the many observed stimulus in same time have same semantic information. This relationship means the semantic concept of linguistic symbol. Moreover, this study proposes the simple Man-machine communication model, to realize the flexible understanding system that is independent of recognition target.
| Translated title of the contribution | Knowledge Acquisition through Man-machine Communication with Image and Voice Media |
|---|---|
| Original language | Japanese |
| Pages (from-to) | 41 - 48 |
| Journal | IEICE technical report. Natural language understanding and models of communication |
| Volume | 97 |
| Issue number | 594 |
| State | Published - 13 Mar 1998 |