Dr. Hsu's research interests include data science, macro-informatics, cognitive computing, interconnection networks, AI, and combinatorial fusion analysis(CFA). His group has been exploring various aspects of combinatorial fusion including its
architecture, algorithms, and applications since 2002 when it was proposed in a DIMACS Technical Report (Hsu, Shapiro & Taksa; 2002-58 (2002), p.1-47). Methods and practices of CFA have been used in a variety of fields in science, technology, and society. These include (among others): biomedical informatics and cheminformatics, brain science and computational social science, cognitive neuroscience and decision making, information technology and wireless networking, financial informatics and business intelligence, machine learning and multi-layer combinatorial fusion.
Among the results achieved by Dr. Hsu's group in the Laboratory of Informatics and Data Mining (LIDM) are: ChIP-seq peak detection, target tracking with occlusion, consensus scoring in virtual screening, stress identification, wireless network selection, market segmentation, portfolio management, intrusion detection, gaze preference prediction using eye movement, and joint decision making.
One of the current research focuses is combinatorial learning theory (CLT). In particular, we are building the foundation of and framework for Deep Fusion which combines deep learning with combinatorial fusion. The Bubble Sort Cayley graph on the symmetric group of order n is used as the learning space with n factorial nodes where n is the number of data items under consideration. CLT will provide efficient and effective learning algorithms to achieve the goal of each of the domain applications in multiple regression, classification, ensemble methods, and other machine learning techniques.