In this paper we first introduce four kinds of modification of Symmetric Scoring [1] which produce likelihood ratios that do not need to be explicitly normalized, i.e. T-norm, Z...
The notion of using context information for solving highlevel vision problems has been increasingly realized in the field. However, how to learn an effective and efficient context...
Abstract. Speeding up approximate pattern matching is a line of research in stringology since the 80’s. Practically fast approaches belong to the class of filtration algorithms,...
We propose a new method for solving structured CSPs which generalizes and improves the Cyclic-Clustering approach [4]. First, the cutset and the tree-decomposition of the constrai...
— Many image compression techniques require the quantization of multiple vector sources with significantly different distributions. With vector quantization (VQ), these sources ...