Abstract

The particle filter (PF) algorithms, also known as Sequential Monte Carlo (SMC) methods, have revolutionized probabilistic state filtering for dynamic systems, while in this talk, Dr. Liu will show how to use PF to solve typical numerical optimization problems that appeared in different communities such as evolutionary computation (EC) and machine learning (ML). Specifically, a posterior exploration based SMC method for derivative-free optimization and a series of PF-type incremental proximity methods for large-scale stochastic optimization will be introduced. Both pros and cons of such PF-type optimization methods will be discussed.

Dr. Bin Liu received the B.E. degree and the Ph.D. degree from Beijing Univ. of Posts and Telecommunications in 2004 and the Chinese Academy of Sciences in 2009, respectively. Then he held a position of research scholar with the Department of Statistical Science at Duke University and was simultaneously appointed as a research scholar with the Statistical and Applied Mathematical Sciences Institute (SAMSI), in the period from 2009 to 2010. He was a year-long visiting scholar at Department of Statistics and Center for the Neural Basis of Cognition in Carnegie Mellon University, and a short-term visiting scholar at the Department of Computer Science at the University of Surrey. He is now an Associate Professor with School of Computer Science, Nanjing University of Posts and Telecommunications (NUPT). His scientific interests span from Bayesian inference (especially Monte Carlo methods), statistical modeling, machine learning to stochastic optimization. He was the recipient of the Best Paper Award of the 10th International Conference on Advanced Computational Intelligence (ICACI 2018), the Scientific and Technological Prize of NUPT and the Best Presentation Award of the 2nd International Conference on Mechanical, Automotive and Materials Engineering (CMAME 2014).