A probability distribution kernel based on whitening transformation

A kernel function can change linearly inseparable vectors into linearly separable vectors by mapping the original feature space into another feature space. But the dimension of the new feature space is usually several times of the original feature space, which results in a more computational complexity. This paper aims at introducing a Dirichlet probability distribution kernel based on whitening transformation which is called DPWT kernel function for short. The DPWT kernel function first mapping feature vectors of the original feature space into new vectors of another same dimension feature space, and then classifies new vectors in the new feature space so as to achieve the purpose of classifying original feature vectors. The DPWT kernel doesn’t augment the dimension of the new feature space. What’s more, the DPWT kernel can effectively eliminate the correlation between vectors, and reduce the redundancy of data which can further improve the accuracy of classification. In this paper, we use the DPWT kernel and other five commonly used kernels on the three benchmark datasets (VOC2007, UIUCsport and Caltech101) for image classification experiments. Experiments show that the DPWT kernel exhibits superior performances compared to the other state of the art kernels.

[1]  C. V. Jawahar,et al.  Generalized RBF feature maps for Efficient Detection , 2010, BMVC.

[2]  Radu Tudor Ionescu,et al.  Kernels for Visual Words Histograms , 2013, ICIAP.

[3]  Andrew Zisserman,et al.  Efficient Additive Kernels via Explicit Feature Maps , 2012, IEEE Trans. Pattern Anal. Mach. Intell..

[4]  Subhransu Maji,et al.  Efficient Classification for Additive Kernel SVMs , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[5]  Mustafa Neamah Jebur,et al.  Spatial prediction of landslide hazard at the Luxi area (China) using support vector machines , 2015, Environmental Earth Sciences.

[6]  Michael G. Madden,et al.  The Genetic Kernel Support Vector Machine: Description and Evaluation , 2005, Artificial Intelligence Review.

[7]  Subhransu Maji,et al.  Classification using intersection kernel support vector machines is efficient , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[8]  P. Szabó.  Response to “Variable directionality of gene expression changes across generations does not constitute negative evidence of epigenetic inheritance” Sharma, A. Environmental Epigenetics, 2015, 1-5 , 2016, Genome Biology.

[9]  Gang Wang,et al.  Learning image similarity from Flickr groups using Stochastic Intersection Kernel MAchines , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[10]  Hwann-Tzong Chen,et al.  A square-root sampling approach to fast histogram-based search , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[11]  Cordelia Schmid,et al.  Multimodal semi-supervised learning for image classification , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[12]  Mohammad H. Mahoor,et al.  Task-dependent multi-task multiple kernel learning for facial action unit detection , 2016, Pattern Recognit..

[13]  Wen Gao,et al.  Group-sensitive multiple kernel learning for object categorization , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[14]  Prabir Bhattacharya,et al.  Function Dot Product Kernels for Support Vector Machine , 2006, 18th International Conference on Pattern Recognition (ICPR'06).

[15]  Andrew Zisserman,et al.  The devil is in the details: an evaluation of recent feature encoding methods , 2011, BMVC.

[16]  Ahmed M. Elgammal,et al.  Learning Kernels for Structured Prediction using Polynomial Kernel Transformations , 2016, ArXiv.

[17]  Florent Perronnin,et al.  Large-scale image categorization with explicit data embedding , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[18]  Benjamin Recht,et al.  Random Features for Large-Scale Kernel Machines , 2007, NIPS.

[19]  Di Chen,et al.  Relative Error Embeddings of the Gaussian Kernel Distance , 2016, ALT.