—Beyond conventional linear and kernel-based feature extraction, we present a more generalized formulation for feature extraction in this paper. Two representative algorithms usi...
Beyond conventional linear and kernel-based feature extraction, we present a more generalized formulation for feature extraction in this paper. Two representative algorithms using ...
The foremost nonlinear dimensionality reduction algorithms provide an embedding only for the given training data, with no straightforward extension for test points. This shortcomin...
Abstract. Multiple-instance learning (MIL) allows for training classifiers from ambiguously labeled data. In computer vision, this learning paradigm has been recently used in many ...
The paper presents results of the face verification contest that was organized in conjunction with International Conference on Pattern Recognition 2000 [14]. Participants had to u...
Jiri Matas, Miroslav Hamouz, Kenneth Jonsson, Jose...