Prof. Byeong Ho Kang

Professor,
Discipline Head of ICT,
University of Tasmania



Title : Seamless knowledge maintenance for Big Data and IoT

Abstract

The computer programmer cannot develop software without knowledge of the target domain. This explains why explicit knowledge domain, such as a patient record management software in hospital, is easier than the implicit knowledge domain, a medical diagnosis system. Internet of Things (IoT) technology is adding more complications into this because connected input/output devices increase complexity of knowledge exponentially in both knowledge domains. The level of software services will be related to the understanding level of relational operation knowledge among multi-mobile devices. Unfortunately, there are not many commercial AI software in industry although these approaches have been attempted for more than two decades. Then there is a question why this approach was not well accepted in industrial applications. In early AI period, there were two approaches, human-driven knowledge generation and data-driven knowledge generation. Data-driven knowledge generation approach known as machine learning has been preferred by AI developers and researchers because human knowledge is implicit and we still don't know how human accumulates their knowledge from their experiences. In this approach, the major bottleneck of AI system in commercial use was a knowledge maintenance issue. The machine learning does not provide a function that can update knowledge on the fly because it is based on statistical modelling on collected data and the streaming of cases to the training sets are not considered. In order to add a new or missing knowledge, the entire knowledge should be regenerated from new data sets. We must discuss why the knowledge maintenance is not easy in this approach. We propose an efficient hybrid artificial intelligence architecture developed at the University of Tasmania and University of New South Wales. The system proposes the integration of both data-driven and human-driven approaches, Multiple Classification Ripple Down Rules by Byeong Ho Kang and Paul Compton and Induct by Brian Gains. With various successful commercial cases, we recently integrated two different approaches together. In any situation, the system can start with one of these approaches and incrementally maintain knowledge bases. We are experimenting this approach in a network security domain.


Biography

Professor Byeong Ho Kang is the Head of Discipline of Information, Communication and Technology, University of Tasmania and he has worked in the University of Tasmania since 2000. Over his 25-year professional career, he worked in the Advanced Research Lab HITACHI in Japan, and Kyung Hee University in Korea since he received his Ph.D from the University of New South Wales, Australia, in 1996. He hasalso taken part in major research and development projects funded by industries and research organizations including the Smart Internet Collaborative Research Centre, the US Air Force, US Navy, Hyundai Motor Groupand many others. Also, he has led and participated in several major industrial funding programs by Ministries of Australian and Korean governments.

His research interests include basic knowledge acquisition methods and applied researches in internet systems as well as medical expert systems. He has been involved in the development of several commercial and internet based applications including AI products, expert system development tools, intelligent help desk systems, Web based information monitoring and classification systems and many more. In addition, he has played a major role in the foundation of several spinning companies.

Professor Kang has published more than 250 papers in journals and international conferences, and was an editor for several books of proceedings and has been a guest editor for special issues in journals.He has served as a chair and steering committee member of many international organizations and conferences. His skills are clearly illustrated in the various leadership roles he takes. He was the chair of Australian Artificial Intelligent Conference in 2006 and 2016, and has led the community of Pacific Knowledge Acquisition area. He is a steering committee member of Pacific Rim International Conference on AI and Australian AI conference and is also currently serving as President of Korean Academy of Scientists and Engineers in Australia.