In kernel methods, such as support vector machines, many existing kernels consider similarity between data by taking into account only their content and without context. In this paper, we propose an alternative that upgrades and further enhances usual kernels by making them context-aware. The proposed method is based on the optimization of an objective function mixing content, regularization and also context. We will show that the underlying kernel solution converges to a positive semi-definite similarity, which can also be expressed as a dot product involving "explicit" kernel maps. When combining these context-aware kernels with support vector machines, performances substantially improve for the challenging task of image annotation.
展开▼
