Context-Based Support Vector Machines for Interconnected Image Annotation

We introduce in this paper a novel image annotation approach based on support vector machines (SVMs) and a new class of kernels referred to as context-dependent. The method goes beyond the naive use of the intrinsic low level features (such as color, texture, shape, etc.) and context-free kernels, in order to design a kernel function applicable to interconnected databases such as social networks. The main contribution of our method includes (i) a variational approach which helps designing this function using both intrinsic features and the underlying contextual information resulting from different links and (ii) the proof of convergence of our kernel to a positive definite fixed-point, usable for SVM training and other kernel methods. When plugged in SVMs, our context-dependent kernel consistently improves the performance of image annotation, compared to context-free kernels, on hundreds of thousands