We are pleased to have Dr. Bin Liu, Director of the Research Centre of Applied Mathematics and Machine Intelligence at Zhejiang Lab, join us for this ART-AI seminar, entitled ‘Uncertainty Modelling & Inference with Applications From Bayesian Computation to Deep Learning’. This seminar will take place on Zoom on Wednesday 3rd November 10.00am-11.00am. ART-AI student Elsa Zhong will chair. If you would like to join, please e-mail [email protected].
In this talk, Dr. Bin Liu will give a summary of one major line of his research experience that ranges from Bayesian computation to deep learning. The content of this talk consists of three parts;
- Bayesian computation (in particular, adaptive annealed importance sampler, a type of Sequential Monte Carlo method)
- The interface between Bayesian computation and machine learning (especially the Bayesian dynamic multi-model assembling theory, which is proposed and developed by Dr. Bin Liu himself in the last 10 years)
- Deep learning
This talk shows the power of uncertainty modelling and qualification, which can be implemented via Bayesian computation, for robust online inference. Algorithms derived based on the theory and their broad applications in extrasolar planet detection, neuro-decoding in brain-computer interface, and label-free CNN model selection are briefly discussed.
Dr. Bin Liu is Director of the Research Centre of Applied Mathematics and Machine Intelligence at Zhejiang Lab. He received a Ph.D. degree in signal and information processing from the Chinese Academy of Sciences in January 2009. He was a research scholar at the Department of Statistical Science, Duke University, a research project member at Statistical and Applied Mathematical Sciences Institute, a visiting scholar at CMU, a senior research fellow at Kuang-Chi Institute of Advanced Technology, an associate professor at Nanjing Univ. of Posts and Telecom., and a senior algorithm expert at Alibaba Group. His research has been in the area of Bayesian statistics, machine learning, and signal processing, with primary interests in the theory of uncertainty modelling; robust inference & learning in dynamic systems; optimization; and applications of the theory in a wide range of disciplines. He has published 60+ papers in refereed journals and conferences like IEEE Trans, NeurIPS, ICASSP, SPL, Knowledge-Based Systems, Neurocomputing, Astrophysical Journal Supplement Series and Journal of Global Optimization.