Thời gian: thứ 6, ngày 09/08/2019
– Buổi sáng: 9:00 – 11:30
– Buổi chiều: 13:30 – 16:00
Địa điểm: Phòng 304, nhà G2, Số 144, Xuân Thủy, Cầu Giấy
Tiêu đề: Adaptive filtering and its application to high-dimensional systems
Giảng viên: TS. Hoàng Hồng Sơn, French Institute for Marine Research in Hydrography and Oceanography
Toulouse, France
ABSTRACT
The inverse engineering problems approach is a discipline that is growing very rapidly. The
inverse problems we talk about in this lecture concern the way to determine the state and/or
parameters of the physical system of interest using observed measurements. In this context the
filtering algorithms like the famous Kalman filter (KF)
constitute a key tool to offer improvement
of our knowledge on the system state, its forecast… which are essential, in particular, for very
high dimensional systems: operational prediction systems in geophysical systems (meteorology,
oceanography)
; image processing; visualizing data; power systems; psychology: children test
results (allergens…)
; finance: study of volatility dynamics; image processing: from eigenfaces
to deep learning (identification of human face…)
; genomics: DNA microarray data;… Many modern
applications fall into the “ultra-high dimension” case with much more variables than observations
(p << n)
like operational prediction systems in geophysical 8 5 systems (meteorology,
oceanography, image processing: n ~ 10 ; p ~ 10 )
. The objective of this lecture is to give an
overview on how one can design a simple, no time-consuming Adaptive Filter (AF)
to solve the
inverse engineering problems with high forecasting performance in very high dimensional
environment with uncertainties. Numerous examples will be given to illustrate the AF approach and
its comparison, especially with two widely used approaches: Kalman filtering and Variational data
assimilation.
BIOGRAPHY
Dr. Hoang Hong Son was born on July 12, 1954 in Vietnam. He has been working for SHOM/LEGOS,
Toulouse, France as a researcher and a consultant since 1993. His main researches contain control
and filtering in stochastic systems, adaptive filters for very high dimensional systems and
dynamical oceanography. He had bachalor degree in Mathematics, Belorussian State University,
Minsk, Belarus, 1977 and Ph.D. degree in Applied Mathematics, Polytechnic Institute, Hanoi,
Vietnam, 1988. He aslo was a researcher in Institute Informatics, Hanoi, Vietnam, 1977 – 1979,
Institut Defense Technology, Hanoi, 1980 – 1983, Institute Physics, Hanoi, 1984 – 1991. He also
was a visiting researcher in LMD/École Normale Supérieure, Paris, 1991 – 1993. Dr. Hong Son Hoang
has been listed as a noteworthy mathematician, researcher by Marquis.