While recent techniques for discriminative dictionary learning have demonstrated tremendous success in image analysis applications, their performance is often limited by the amount of labeled data available for training. Even though labeling images is difficult, it is relatively easy to collect unlabeled images either by querying the web or from public datasets. Using the kernel method, we prop...

For a poset P and a distributive 〈∨, 0〉-semilattice S, a S-valued poset measure on P is a map μ : P×P → S such that μ(x, z) ≤ μ(x, y)∨μ(y, z), and x ≤ y implies that μ(x, y) = 0, for all x, y, z ∈ P . In relation with congruence lattice representation problems, we consider the problem whether such a measure can be extended to a poset measure μ : P ×P → S, for a larger poset P , such that for al...

Computers have already eclipsed the level of human play in competitive Scrabble, but there remains room for improvement. In particular, there is much to be gained by incorporating information about the opponent’s tiles into the decision-making process. In this work, we quantify the value of knowing what letters the opponent has. We use observations from previous plays to predict what tiles our ...

Interactive POMDPs (I-POMDPs) are a useful framework for describing POMDPs that interact with other POMDPs. I-POMDPs are solved recursively in levels: a level1 I-POMDP assumes the opponent acts randomly, and a levelk I-POMDP assumes the opponent is a level-(k-1) I-POMDP. In this paper, we introduce fully-nested I-POMDPs, which are uncertain about the physical state of the game, the level of the...

In ci componioiz paper [l] we have developed a frame,work fo,r rating 01‘ comparing navigation packages. For a giuen enuir.onment a navigation package consists of (I motion planner and a sensor to be used during nauigation. The cibility to rate or measure a navigation package is aniportant in order to address issues like sensor custom%zution for an environment and choice of a m.otion plarumr in...

in Euclidean spaces, with Fn(x, a ) and Gn(x) converging pointwise to functions F,(x,a) and G,(x), respectively, and give conditions for the limiting P O model Xt+l = F,(xt,at) + t t , Yt = G,(xt) + rlt to have an a-discount optimal policy. AMS Classification: 93320, 90C40.

The partially observable Markov decision process (POMDP) framework has been applied in dialogue systems as a formal framework to represent uncertainty explicitly while being robust to noise. In this context, estimating the dialogue POMDP model components is a significant challenge as they have a direct impact on the optimized dialogue POMDP policy. To achieve such an estimation, we propose meth...

We consider a special case of partially observable Markov decision processes that arises when state information is perfect but arrives with a delay. We rst formulate the decision process in its standard form and derive the Bellman equation that corresponds to it. We then introduce a second decision process that has a much simpler Bellman equation than the rst, and is therefore, in general, much...

We study the problem of finding an optimal policy for a Partially Observable Markov Decision Process (POMDP) when the model is not perfectly known and may change over time. We present the algorithm MEDUSA+, which incrementally improves a POMDP model using selected queries, while still optimizing the reward. Empirical results show the response of the algorithm to changes in the parameters of a m...

We discuss the importance of using partially quenched theories with three degenerate quarks for extrapolating to QCD, and present some relevant results from chiral perturbation theory. Simulations of lattice QCD have, to date, largely used the quenched (Q), or partially quenched (PQ), approximations. These approximations introduce unphysical artifacts, some of which can be analyzed using chiral...