Biography: Lei Zhang received his Ph.D degree in Circuits and Systems from the College of Communication Engineering, Chongqing University, Chongqing, China, in 2013. He was selected as a Hong Kong Scholar in China in 2013, and worked as a Post-Doctoral Fellow with The Hong Kong Polytechnic University, Hong Kong, from 2013 to 2015. He is currently a Professor/Distinguished Research Fellow with Chongqing University. He has authored more than 60 scientific papers in top journals, including the IEEE TRANSACTIONS ON IMAGE PROCESSING, the IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, the IEEE TRANSACTIONS ON MULTIMEDIA, the IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS, the IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, the IEEE SENSORS JOURNAL, INFORMATION FUSION, SENSORS & ACTUATORS B, and ANALYTICA CHIMICA ACTA. His current research interests include machine learning, pattern recognition, computer vision and intelligent systems. Dr. Zhang was a recipient of Outstanding Reviewer of Emerald, in 2016, Outstanding Doctoral Dissertation Award of Chongqing, China, in 2015, Hong Kong Scholar Award in 2014, Academy Award for Youth Innovation of Chongqing University in 2013 and the New Academic Researcher Award for Doctoral Candidates from the Ministry of Education, China, in 2012.
Speech Title: New Solutions for Non-i.i.d. Data Mining: Transfer Learning
Abstract: Machine learning plays an increasing important role in artificial intelligence, computer vision, and data mining, etc. The key lies in that the training set and testing set are with independent and identical distribution (i.i.d.). However, this assumption does not hold with the increasing scale of big data, which results in the performance degradation of conventional recognition/classification methods. In this topic, I will introduce several new Transfer Learning methods for addressing the data non-i.i.d. problem. The existing transfer learning methods will be first reviewed, and then I will share our new transfer learning methods, such as Latent Sparse Domain Transfer (LSDT), Extreme Domain Adaptation (EDA) and Discriminative Kernel Transfer Learning (DKTL) for domain adaptation and visual categorization of domain data. The model and optimization algorithms will be clearly presented. Finally, the experiments on benchmark datasets with image classification, face recognition, handwritten digits recognition and object recognition will be shown. Some new insights on transfer learning will be discussed. More information can be referred to our website: http://www.escience.cn/people/lei/index.html