CORMSIS Seminar Event
- Time:
- 16:00 - 17:00
- Date:
- 27 November 2014
- Venue:
- TBA
For more information regarding this event, please telephone Houdou Qi on +23683 or email H.Qi@soton.ac.uk .
Event details
Metric Learning with Eigenvalue Optimization
Abstract:
Distance metric is a fundamental concept in Machine Learning since a proper choice of a metric has crucial effects on the performance of both supervised and unsupervised learning algorithms. In this talk I will present our recent work in this challenging research direction, starting with an introduction to metric learning problems. Firstly, I will introduce a novel metric learning approach (DML-eig) which is shown to be equivalent to a well-known eigenvalue optimization problem called minimizing the maximal eigenvalue of a symmetric matrix. Moreover, I will show that similar ideas can be extended to other learning problems such as large margin nearest classifiers (LMNN) and maximum-margin matrix factorisation for collaborative filtering (e.g. predicting customers’ preference to products). Then, first-order algorithms scalable to large datasets are developed and their convergence analysis will be discussed. At last, the competitiveness of our methods are shown by various experiments on benchmark datasets. In particular, I will report some encouraging results on a challenging face verification dataset called Labeled Faces in the Wild (LFW).
Speaker information
Dr Yiming Ying ,University of Exeter,Dr. Ying received a BSc (1997) in Mathematics from Zhejiang University (formally Hangzhou University) and a PhD (2002) in Mathematics from Zhejiang University, China. He is a lecturer in computer science at the University of Exeter since 2010. Before that, he has been a postdoctoral researcher at the City University of Hong Kong until 2005 working on learning theory, at the University College London until 2007 working on theoretical machine learning, and at the University of Bristol specialising in machine learning and applications to cancer informatics. His research interests include machine learning, learning theory, optimization for big data, and applications to computer vision and bioinformatics.