Recent Publications and Preprints
Full list of publications on both Statistical Machine Learning and Mathematics (Harmonic Analysis),
see my CV
Preprint
- Y. Ying and D. X. Zhou. Online pairwise learning algorithms. Preprint, 2014.
- Q. Cao, Y. Ying, M. Pontil, and Peng Li.
Similarity metric learning for unconstrained face verification and person re-identification.
Submitted for publication, 2014.
- Y. Lei and Y. Ying. Generalization analysis for multi-modal metric learning. Submitted for publication, 2014.
- M. Rogers, C. Campbell and Y. Ying. Probabilistic inference of biological networks via data integration. To appear in BioMed Research International, 2014.
2014
2013
2012
2011
2010
This second reversion is a substantial extension of the COLT (2009) conference paper: Generalization bounds for learning the kernel. In particular, we provided a self-contained proof for bounding the Rademacher chaos complexity by metric entropy integrals and also corrected the inaccurate claim on generalization bounds derived from the covering number approach (appeared at the end of Section 3 of the COLT conference version).
2009
2008
MATLAB code available under request.
2007
- A. Caponnetto, C. A. Micchelli, M. Pontil, and Y. Ying, Universal multi-task kernels,
Journal of Machine Learning Research, 9 (2008), 1615-1646. (Was Technical Report, University College London, December 2006)
Characterization of universal matrixed-valued (and more general operator-valued) kernels.
These characterizations are highlighted with numerous examples of paractical importance in multi-task learning.
2006
Generalization analysis of Online Stochastic Gradient Descent algorithms in reproducing kernel Hilbert space.
In particular, we show that their error rates are competitive with offline regularization algorithms.
2005
Characterization of statistical consistency for learning the kernel algorithms
by the V-gamma dimension of the set of candidate kernels.
2004
Return to Yiming Ying's home page.