Online Learning Algorithms in Machine Learning

Speaker:        Dr. Yiming Ying
                Department of Mathematics and Statistics
                State University of New York at Albany
                USA

Title:          "Online Learning Algorithms in Machine Learning"

Date:           Wednesday, 15 June 2016

Time:           2:00pm - 3:00pm

Venue:          Room 1511 (near lifts 27/28), HKUST

Abstract:

Many machine learning tasks can be formulated as the problem of minimizing
an objective function defined by a sum of losses over the training data.
The main difficulty in obtaining optimal solutions of such formulations is
the big volume of the fast-updating data (large n and p). Online learning
algorithms, which pass over the data only once, are widely used in
practice. In this talk, I will review some of our theoretical
contributions to this important research direction. In particular, for
pointwise learning problems such as classification I will review
convergence results of online learning algorithms in a general setting of
RKHS for both regularized and un-regularized formulations. For pairwise
learning problems such as AUC maximization/bipartite ranking, I will show
its equivalence to a stochastic saddle point problem. From this equivalent
formulation, a truly online algorithm for AUC maximization will be
developed.

***********************
Biography:

Yiming Ying received the B.S. and PhD degrees in Mathematics, respectively
in 1997 and 2002, from Zhejiang University, Hangzhou, China. He then
worked as a post-doctoral researcher at the City University of Hong Kong,
University College London and University of Bristol before he became a
Lecturer in Computer Science at the University of Exeter (UK) in 2010. In
2015, he moved to State University of New York at Albany (USA) where he is
currently an Associate Professor in the Department of Mathematics and
Statistics. His research interests include learning theory, machine
learning, large-scale optimization and applications to computer vision and
bioinformatics.