Search

In this paper we investigate systematic sampling in the image- synthesis context. Systematic sampling has been widely used in stereology to improve the efficiency of different probes in experimental design. These designs are theoretically based on estimators of 1-dimensional and 2-dimensional integrals. For the particular case of the characteristic function, the variance of these estimators has been shown to be asymptotically N -3/2, which improves on the O(N -1) behaviour of independent estimators using uniform sampling. Thus, when no a priori knowledge of the integrand function is available, like in several image synthesis techniques, systematic sampling efficiently reduces the computational cost. [ABSTRACT FROM AUTHOR]

In this paper we describe the ICSI-SRI entry in the Rich Transcription 2005 Spring Meeting Recognition Evaluation. The current system is based on the ICSI-SRI clustering system for Broadcast News (BN), with extra modules to process the different meetings tasks in which we participated. Our base system uses agglomerative clustering with a modified Bayesian Information Criterion (BIC) measure to determine when to stop merging clusters and to decide which pairs of clusters to merge. This approach does not require any pre-trained models, thus increasing robustness and simplifying the port from BN to the meetings domain. For the meetings domain, we have added several features to our baseline clustering system, including a "purification" module that tries to keep the clusters acoustically homogeneous throughout the clustering process, and a delay&sum beamforming algorithm which enhances signal quality for the multiple distant microphones (MDM) sub-task. In post-evaluation work we further improved the delay&sum algorithm, experimented with a new speech/non-speech detector and proposed a new system for the lecture room environment. [ABSTRACT FROM AUTHOR]

Adaptive sampling is an interesting tool to eliminate noise, which is one of the main problems of Monte Carlo global illumination algorithms. We investigate the Tsallis entropy to do adaptive sampling. Implementation results show that adaptive sampling based on Tsallis entropy consistently outperforms the counterpart based on Shannon entropy. [ABSTRACT FROM AUTHOR]

Context unification (CU) is the open problem of solving context equations for trees. We distinguish a new decidable variant of CU-well-nested CU - and present a new unification algorithm that solves well-nested context equations in non-deterministic polynomial time. We show that minimal well-nested solutions of context equations can be composed from the material present in the equation (see Theorem[1]). This property is wishful when modeling natural language ellipsis in CU. [ABSTRACT FROM AUTHOR]

The constraint language for lambda structures (CLLS) is a description language for lambda terms. CLLS provides parallelism constraints to talk about the tree structure of lambda terms, and lambda binding constraints to specify variable binding. Parallelism constraints alone have the same expressiveness as context unification. In this paper, we show that lambda binding constraints can also be expressed in context unification when permitting tree regular constraints. Keywords: second-order unification, dominance constraints, computational linguistics, underspecified semantics. [ABSTRACT FROM AUTHOR]

The fitness of human immunodeficiency virus type-1 (HIV-1) appears to be an important piece of the puzzle in understanding the global HIV-1 epidemic and disease progression in infected individuals. We have developed a dual infection/competition assay followed by a sensitive heteroduplex tracking assay (HTA) to measure the relative fitness of any HIV-1 isolate. Differences in fitness between wild-type and drug-resistant HIV-1 isolates can also be used as indicators of the emergence of drug-resistant isolates. This chapter describes the methods utilized to measure viral fitness (relative replicative capacity) during growth competition experiments. HTA is used to evaluate the production of HIV-1 variants in a competition, which can then be used to estimate ex vivo viral fitness. [ABSTRACT FROM AUTHOR]

Linear Second-Order Unification and Context Unification are closely related problems. However, their equivalence was never formally proved. Context unification is a restriction of linear second-order unification. Here we prove that linear second-order unification can be reduced to context unification with tree-regular constraints. Decidability of context unification is still an open question. We comment on the possibility that linear second-order unification is decidable, if context unification is, and how to get rid of the tree-regular constraints. This is done by reducing rank-bound tree-regular constraints to word-regular constraints. [ABSTRACT FROM AUTHOR]

Our goal is to characterize and to be able to predict the search cost, of some of the most important CSP algorithms and heuristics when solving CSP problems by obtaining a statistical model of the algorithm runtime based on inexpensively computed parameters obtained from the CSP problem specification and the associated constraints and nogoods graphs. Such a model will give us three important items concerning the studied CSP problems. First, the model provides a tool to predict the search cost of a given instance, allowing a portfolio of solvers to decide for the best algorithm before to proceed. Second, the models will give an insight about which are the main features that characterize the complexity of a RBCSP. Finally, another potential benefit of the model is pointing out which features are the algorithms most sensible to, thus helping to guess potential areas of improvement. [ABSTRACT FROM AUTHOR]