Kernel Learning and Data Integration

 

 

 

Kernel methods have been extensively used for modern data analysis. They can incorporate discriminatory features into kernel matrices which encode the similarity between samples in respective data sources. The performance of a kernel machine largely depends on the data representation via the choice of kernel function. Hence, one central issue in kernel methods is the problem of kernel selection. Kernel learning can range from the width parameter selection of Gaussian kernels to obtaining an optimal linear combination from a set of finite candidate kernels. The latter is often referred to as multiple kernel learning (MKL) in Machine Learning and non-parametric Group Lasso in Statistics.

 

 

 

  Description: C:\Users\yy267\yiming\private\ML-group-design\mkl.png                                          Description: C:\Users\yy267\yiming\private\ML-group-design\yeast.png

 

 

 

Firstly, the theoretical part of this project is to address the statistical and computational challenges of kernel learning problems such as generalization analysis and the design efficient algorithms for massive data sources.  Secondly, perhaps more importantly, we apply these theoretical principles to integrating multiple biomedical datasets to enhance the biological inference and fusing various image descriptors in face verification.

 

Selected Publications:

 

1.    Y. Ying, K. Huang and C. Campbell, Enhanced protein fold recognition through a novel data integration approach, BMC Bioinformatics (Open access), (2009) 10:267.

2.    Y. Ying and D.X. Zhou, Learnability of Gaussians with flexible variances , Journal of Machine Learning Research, 8 (2007), 249-276.

3.    Y. Ying and C. Campbell, Rademacher chaos complexity for learning the kernel, Neural Computation, Vol. 22 (11), 2010.

4.    Y. Ying and C. Campbell, Generalization bounds for learning the kernel, Proceedings of the 22nd Annual Conference on Learning Theory (COLT), 2009.

5.    T. Damoulas, Y. Ying, M. Girolami, and C. Campbell, Inferring sparse kernel combination and relevance vectors: an application to subcelluar localization of proteins , International Conference on Machine Learning and Applications (ICMLA), 2008.

6.    Q. Wu, Y. Ying and D.X. Zhou, Multi-kernel regularized classifiers, Journal of Complexity, 2006. (Was preprint, 2004)

 

 

Return to Yiming Ying's home page