Sungyoung Lee

Also published under: S. Y. Lee, Sung Young Lee, SungYoung Lee, Sungyoun Lee, Sungyoung Y. Lee


Department of Computer Science and Engineering
Kyung Hee University
Gyeonggi-do, Korea
Deep learning and machine learning technologies are being used to solve specific problems by deriving models that approximate real phenomena based on a limited amount of data. However, these technologies have a simple structure than humans that utilize inductive, deductive, and semantic reasoning methods. In particular, the current deep artificial neural network technology simulates the activity of neurons and synapses, which is suitable for simple classification and generation problems, but not suitable for structural and logical reasoning. In order to solve more complex problems, the artificial intelligence algorithms simulate human reasoning and cognitive mechanisms to enable multilayered inferencing. It is necessary to systematically develop an artificial intelligence module that performs human-like logical reasoning by synthesize the results of neural inferencing and symbolic reasoning.

The medical industry mainly uses explainable rule-based symbolic inferencing for reliable decision making, but the expert knowledge acquisition in terms of IF-THEN rules lacks continuous scalability due to high data transformation cost. The recent neural inferencing techniques have solved the knowledge transfer/expandability even in data-deficient domains (e.g., self-supervised learning, transfer-learning), but their full application in the healthcare industry is still limited due to reliability of learned knowledge. Therefore, the neural-symbolic reasoning is required that resolves the complex problems and supports AI algorithms in terms of transparency and high-level interpretability for reliable decision-making. It extends the cognitive mechanism and mimic human-like reasoning for conventional neural learning techniques, which includes the ability to learn and reason from the environmental constraints by producing an appropriate fact for a given decision.

The promising application area of explainable AI and neural-symbolic reasoning includes the analysis of complex networks, social robotics, and health informatics. Overall, neural-symbolic integration seems suitable in application areas where large amounts of heterogeneous data are available and knowledge descriptions are needed. Such approach has earned potential achievements in visual question answering, including robot navigation, health, genomics, hardware/software specification, multimodal data fusion for information retrieval, big data analysis and language understanding


Sungyoung Lee (Member, IEEE) received the B.S. degree from Korea University, Seoul, South Korea, and the M.S. and Ph.D. degrees in computer science from the Illinois Institute of Technology, Chicago, IL, USA, in 1987 and 1991, respectively. He was an Assistant Professor with the Department of Computer Science, Governors State University, University Park, IL, USA, from 1992 to 1993. He has been a Professor with the Department of Computer Engineering, Kyung Hee University, South Korea, since 1993, where he has been the Director of the Neo Medical Ubiquitous-Life Care Information Technology Research Center, since 2006. He is currently the Founding Director of the Ubiquitous Computing Laboratory. His current research interests include ubiquitous computing and applications, wireless ad hoc and sensor networks, context-aware middle-ware, sensor operating systems, real-time systems and embedded systems, and activity and emotion recognition. He is a member of ACM.

(Based on document published on 30 July 2021).
Nha Trang-Vietnam