To improve the precision of top-ranked images returned by a web image search engine, we propose in this paper a novel pseudo relevance feedback method named iterative probabilistic one-class SVMs to re-rank the retrieved images. By assuming that most top-ranked images are relevant to the query, we iteratively train one-class SVMs, and convert the outputs to probabilities so as to combine the de...

We propose a flexible procedure for large-scale image search by hash functions with kernels. Our method treats binary codes and pairwise semantic similarity as latent and observed variables, respectively, in a probabilistic model based on Gaussian processes for binary classification. We present an efficient inference algorithm with the sparse pseudo-input Gaussian process (SPGP) model and paral...

Abstract. We consider a class of nonlinear partial-differential equations, including the spatially homogeneous Fokker-Planck-Landau equation for Maxwell (or pseudo-Maxwell) molecules. Continuing the work of [6, 7, 4], we propose a probabilistic interpretation of such a P.D.E. in terms of a nonlinear stochastic differential equation driven by a standard Brownian motion. We derive a numerical sch...

In this paper we develop a novel probabilistic generalization bound for learning the kernel problem. First, we show that the generalization analysis of the regularized kernel learning system reduces to investigation of the suprema of the Rademacher chaos process of order two over candidate kernels, which we refer to as Rademacher chaos complexity. Next, we show how to estimate the empirical Rad...

An approach to automatic detection of syllable structure is presented. We demonstrate a novel application of EM-based clustering to multivariate data, exempliied by the induction of 3-and 5-dimensional probabilistic syllable classes. The 3-dimensional models were subjected to a pseudo-disambiguation task, the result of which shows that the onset is the most variable, or least predictable, part ...

Discovery of graphical models is NP-hard in general, which justifies using heuristics. We consider four commonly used heuristics. We summarize the underlying assumptions and analyze their implications as to what graphical models they can and cannot discover. In particular, we consider discovery of pseudo-independent (PI) models, a subclass of probabilistic models where subsets of a set of colle...

Relevance feedback has already proven its usefulness in probabilistic information retrieval (IR). In this research we explore whether a pseudo relevance feedback technique on IR can improve the Question Answering task (QA). The basis of our exploration is the use of relevant named entities from the top retrieved documents as clues of relevance. We discuss two interesting findings from these exp...

In this paper, we introduce the  probabilistic normed groups. Among other results, we investigate the continuityof inner automorphisms of a group and the continuity of left and right shifts in probabilistic group-norm. We also study midconvex functions defined  on probabilistic normed groups and  give  some results about locally boundedness of such  functions.

This paper concerns theoretical modeling of degradation of signal with OFDM coding caused by pseudo-random nonlinear distortions introduced by an analog-to-digital or digital-to-analog converter. A new quantity, effective number of samples, is defined and used for derivation of accurate expressions for autocorrelation function and the total power of the distortions. The derivation is based on p...

We compare the failure probabilities of ensemble implementations of quantum algorithms which use pseudo-pure initial states, quantified by their polarization, to those of competing classical probabilistic algorithms. Specifically we consider a class algorithms which require only one bit to output the solution to problems. For large ensemble sizes, we present a general scheme to determine a crit...