University of Tasmania, Australia

We have pioneered techniques whereby knowledge-based systems can be built comparatively rapidly and simply by being instructed on a case-by-case basis, analogous to how an apprentice learns on the job. This approach has had considerable and increasing commercial success in facilitating knowledge-based system development, but the apprentice does not exactly and only what it is told. We propose to develop techniques whereby a developing knowledge-based system can extrapolate beyond what it has learned to new situations and is also aware when novel tasks are beyond its competence. This will greatly increase the speed and ease with which knowledge-based systems can be built.

Members (External)

• Assoc. Prof. Byeong Kang