the problem of hypothesis testing with a nuisance parameter is considered. two methods forusing fuzzy knowledge on the nuisance parameter to test hypotheses are suggested. these methods areneither a pure classical nor a pure bayesian approach to hypothesis testing, but rather related to both. afew known examples and their applications, which cannot be studied by the parametric statisticalmethod...

Algorithms for probabilistic inference in Bayesian networks are known to have running times that are worst-case exponential in the size of the network. For networks with a moralised graph of bounded treewidth, however, these algorithms take a time which is linear in the network’s size. In this paper, we show that under the assumption of the Exponential Time Hypothesis (ETH), small treewidth of ...

We study the potential benefits of unlabeled data to classification prediction to the learner. We compare learning in the semi-supervised model to the standard, supervised PAC (distribution free) model, considering both the realizable and the unrealizable (agnostic) settings. Roughly speaking, our conclusion is that access to unlabeled samples cannot provide sample size guarantees that are bett...

One of the earliest solution concepts considered in game theory are saddle points, combinations of actions such that no player can gain by deviating (see, e.g., von Neumann and Morgenstern, 1947). In two-player zero-sum games, every saddle point happens to coincide with the optimal outcome both players can guarantee in the worst case and thus enjoys a very strong normative foundation. Unfortuna...

Most planners constructed up to now are qualitative: they deal with uncertainty by considering all possible outcomes of each plan, without quantifying their relative likelihood. They then choose a plan that deals with the worstcase scenario. However, it is clearly infeasible to plan for every possible contingency. Even beyond the purely computational considerations, planning for highly unlikely...