Combinatorial Fusion Analysis: Quantifying Ensemble Diversity for Robust ML/AI Systems (Delivered in English)
- LecturerDistinguished Professor D. Frank Hsu (Computer & Information Science Fordham University, USA)
Host: De-Nian Yang - Time2024-08-09 (Fri.) 10:00 ~ 12:00
- LocationAuditorium106 at IIS new Building
Abstract
ML/AI-based methods and systems including ensemble learning and modeling have achieved numerous successes across various domain applications. However, these algorithms and models often lack generalization and robustness. Although evidence suggests that diverse ensembles are more general in its applications and more robust to diverse inputs, understanding the role and utility of ensemble diversity has remained an active area of study.
In this talk, we review Combinatorial Fusion Analysis (CFA): a new paradigm for combining multiple scoring systems (MSSs) in computational learning and ensemble modeling. CFA utilizes rank-score characteristic (RSC) function to characterize individual models and cognitive diversity (CD) to perform model selection and combination. We illustrate the power of CFA and the utility of CD by three recent examples: (a) drug discovery (predicting ADMET properties), (b) sentimental analysis (human-centered ML/AI), and (c) intrusion detection (mitigating DoS attack and enhancing information security)
In this talk, we review Combinatorial Fusion Analysis (CFA): a new paradigm for combining multiple scoring systems (MSSs) in computational learning and ensemble modeling. CFA utilizes rank-score characteristic (RSC) function to characterize individual models and cognitive diversity (CD) to perform model selection and combination. We illustrate the power of CFA and the utility of CD by three recent examples: (a) drug discovery (predicting ADMET properties), (b) sentimental analysis (human-centered ML/AI), and (c) intrusion detection (mitigating DoS attack and enhancing information security)
BIO
Dr. Hsu is the Clavius Distinguished Professor of Science and a Professor of Computer and Information Science at Fordham University, New York, NY, USA. He was visiting professor/scholar at Keio University, Taiwan University, University of Paris-Sud and CNRS, MIT, and DIMACS (a NSF-funded research consortium) at Rutgers University. Dr. Hsu graduated from Cheng Kung University (B.S.), University of Texas at El Paso (M.S.), and University of Michigan (Ph.D.).
Dr. Hsu’s research interests include interconnection network and Combinatorial Fusion Analysis (CFA) as foundation of and with applications to computational learning and modeling, X-informatics, multi-model fusion, and intelligent ML/AI systems. He has served on the editorial boards of several journals including as co-founding editor of Journal of Interconnection Networks (JOIN). He is a Life Senior member of IEEE (Computer Society and Computational Intelligence Society) and an executive committee member of IEEE New York Section.
Dr. Hsu’s research interests include interconnection network and Combinatorial Fusion Analysis (CFA) as foundation of and with applications to computational learning and modeling, X-informatics, multi-model fusion, and intelligent ML/AI systems. He has served on the editorial boards of several journals including as co-founding editor of Journal of Interconnection Networks (JOIN). He is a Life Senior member of IEEE (Computer Society and Computational Intelligence Society) and an executive committee member of IEEE New York Section.