Multiple kernel learning from sets of partially matching image features

Abstract: Kernel classifiers based on Support Vector Machines (SVM) have achieved state-ofthe-art results in several visual classification tasks, however, recent publications and developments based on SVM have shown that using multiple kernels instead of a single one can enhance interpretability of the decision function and improve classifier performance, which motivates researchers to explore the use of homogeneous model obtained as linear combinations of kernels. Multiple Kernel Learning (MKL) allows the practitioner to get accurate classification results and identify relevant and meaningful features. However, the use of multiple kernels faces the challenge of choosing the kernel weights, and an increased number of parameters that may lead to overfitting. In this paper we show that MKL problem can be formulated as a convex optimization problem, which can be solved efficiently using projected gradient method. Weights on each kernel matrix (level) are included in the standard SVM empir...

• Type: Conference
• Year: 2008
• Where: ICPR
• Authors: Guo ShengYang, Min Tan, Si-Yao Fu, Zeng-Guang Hou, Zi-ze Liang
• Keywords: Computer Vision, Histogram Intersection Kernels, ICPR 2008, Multiple Kernels, SVM Empirical Risk