A number of natural models for learning in the limit is introduced to deal with the situation when a learner is required to provide a grammar covering the input even if only a par...
Our first objective in participating in this domain-specific evaluation campaign is to propose and evaluate various indexing and search strategies for the German, English and Russ...
Abstract. The computation of the probability of string generation according to stochastic grammars given only some of the symbols that compose it underlies pattern recognition prob...
This paper presents new and effective algorithms for learning kernels. In particular, as shown by our empirical results, these algorithms consistently outperform the so-called uni...
Software-controlled data prefetching is a promising technique for improving the performance of the memory subsystem to match today's high-performance processors. While prefet...