UPM|Mediateca

Prof. Xu Lei, The Chinese University of Hong Kong, EURASC Fellow Symposium Artificial Intelligence and Ceremony of Awards. 21 and 22 October 2019. Insights on learning and intelligence are provided from a deep bidirectional perspective, featured by inward encoding / cognition and outward reconstruction / implementation. First, we make an overview on bidirectional learning from those studied in the later eighties and the early nineties, such as autoencoder, Lmser reconstruction, and BYY harmony learning, to ones developed in recent years, such as variational autoencoders, deep generative models, GAN, U-net, and DenseNet. Then, we proceed to bidirectional intelligence, driven by long term dynamics for parameter learning and short term dynamics for image thinking and rational thinking. Image thinking deals with information flow as if thinking was displayed in the real world, exemplified by typical tasks of bidirectional deep learning, while rational thinking handles symbolic strings, performing uncertainty reasoning and problem solving, exemplified by AlphaGoZero like searching, IBM Watson system, and causal computation. Lei Xu, Emeritus Professor, Chinese University of Hong Kong. Zhiyuan Chair Professor, Shanghai Jiao Tong University (SJTU). Chief Scientist, SJTU AI Research Institute. Director of Neural Computation Research Centre in Brain and Intelligence Science-Technology Institute, Shanghai ZhangJiang National Lab. Received several national and international academic awards, including 1993 National Nature Science Award, 1995 Leadership Award from International Neural Networks Society (INNS) and 2006 APNNA Outstanding Achievement Award. Elected to Fellow of IEEE in 2001. Fellow of intl. Association for Pattern Recognition in 2002 and of European Academy of Sciences (EURASC) in 2003. Published about 100 Journal papers, got about 5500 Web-of-Science citations (over 3900 by top-10 papers with 1319 for top-1 and 119 for 10th). Given over dozens keynote /invited lectures at various international conferences. Served as EIC and associate editors of several academic journals, e.g., including Neural Networks (1995-2016), IEEE Tr. Neural Networks (1994-98). Taken various roles in academic societies, e.g., INNS Governing Board (2001-03), the INNS award committee (2002-03), and the Fellow committee of IEEE Computational Intelligence society (2006-07), and the EURASC scientific committee (2014-17).