A general model is proposed for studying ranking problems. We investigate learning methods based on empirical minimization of the natural estimates of the ranking risk. The empiric...
We propose a new stopping condition for a Support Vector Machine (SVM) solver which precisely reflects the objective of the Leave-OneOut error computation. The stopping condition ...
Application systems often need to react with certain actions whenever some preset conditions are satisfied. In many cases, the evaluation of these conditions takes long time, but...
There has been a recent, growing interest in classification and link prediction in structured domains. Methods such as conditional random fields and relational Markov networks sup...
Abstract Opponent-Model search is a game-tree search method that explicitly uses knowledge of the opponent. There is some risk involved in using Opponent-Model search. Both the pre...
H. H. L. M. Donkers, H. Jaap van den Herik, Jos W....