Showing 1 to 10 of 4502 matching Articles
Results per page:
By
Zhou, Xinhui; Kwan, Chiman; Ayhan, Bulent; Kim, Chanwoo; Kumar, K.; Stern, Richard
Show all (6)
Robust speech recognition in noisy and reverberant conditions is an important research area in recent years. Here we present a comparative study of several spatial speech separation methods. The main performance metric is word error rate (WER) under different signaltonoise ratio (SNR) and reverberant conditions. Extensive simulations showed that one technique known as polyaural processing stood out as the best one.
more …
By
Baumstark, Niklas; Blelloch, Guy; Shun, Julian
5 Citations
Motivated by the observation that FIFObased pushrelabel algorithms are able to outperform highest labelbased variants on modern, large maximum flow problem instances, we introduce an efficient implementation of the algorithm that uses coarsegrained parallelism to avoid the problems of existing parallel approaches. We demonstrate good relative and absolute speedups of our algorithm on a set of large graph instances taken from realworld applications. On a modern 40core machine, our parallel implementation outperforms existing sequential implementations by up to a factor of 12 and other parallel implementations by factors of up to 3.
more …
By
Hausser, Roland
This chapter analyzes the functionality of LAgrammar in analysis and generation. Section 5.1 illustrates the process of incremental pragmatic interpretation during analysis, and explains why constituentstructure analysis seems to have such a strong intuitive basis. Section 5.2 discusses the relation between the syntactic generation of strings and the notion of pragmaticosemantic generation, defined as a mapping from utterance meanings to surfaces. Section 5.3 describes the leftassociative approach to analysis and generation based on the Linear Path Hypothesis. Section 5.4 explains three basic problems of generation, and illustrates possible ways to solve the Extraction Problem and the Connection Problem. Section 5.5 addresses the Choice Problem of generation.
more …
By
Subrahmanian, V. S.; Mannes, Aaron; Sliva, Amy; Shakarian, Jana; Dickerson, John P.
Show all (5)
This chapter describes the methodology and the algorithm used to automatically compute policy options. It provides a mathematical definition of a policy against LeT and then proves the LeT Violence NonEliminability Theorem that shows there is no policy that will stop all of LeT’s terrorist actions. The reason for this is that attacks on holidays are carried out in situations that are inconsistent with situations when LeT carries out other types of attacks. The chapter presents an algorithm to compute all policies (in accordance with the mathematical definition of policy) that have good potential to significantly reduce all types of attacks carried out by LeT (except for attacks on holidays). Readers who do not wish to wade through the technical details can skip directly to Sect. 10.5 which summarizes the results of this chapter.
more …
By
Sycara, Katia
9 Citations
While the Web has emerged as a World Wide repository of digitized information, by and large, this information is not available for automated inference. Two recent efforts, the Semantic Web [1] and Web Services hold great promise of making the Web a machine understandable infrastructure where software agents can perform distributed transactions. The Semantic Web transforms the Web into a repository of computer readable data, while Web services provide the tools for the automatic use of that data. To date there are very few points of contact between Web services and the Semantic Web: research on the Semantic Web focuses mostly on markup languages to allow annotation of Web pages and the inferential power needed to derive consequences, utilizing the Web as a formal knowledge base. Web services concentrate on proposals for interoperability standards and protocols to perform B2B transactions.
more …
By
Johnson, Benjamin; Grossklags, Jens; Christin, Nicolas; Chuang, John
Show all (4)
4 Citations
Motivated attackers cannot always be blocked or deterred. In the physicalworld security context, examples include suicide bombers and sexual predators. In computer networks, zeroday exploits unpredictably threaten the information economy and end users. In this paper, we study the conflicting incentives of individuals to act in the light of such threats.
More specifically, in the weakest target game an attacker will always be able to compromise the agent (or agents) with the lowest protection level, but will leave all others unscathed. We find the game to exhibit a number of complex phenomena. It does not admit pure Nash equilibria, and when players are heterogeneous in some cases the game does not even admit mixedstrategy equilibria.
Most outcomes from the weakesttarget game are far from ideal. In fact, payoffs for most players in any Nash equilibrium are far worse than in the game’s social optimum. However, under the rule of a social planner, average security investments are extremely low. The game thus leads to a conflict between pure economic interests, and common social norms that imply that higher levels of security are always desirable.
more …
By
Kociumaka, Tomasz; Pachocki, Jakub W.; Radoszewski, Jakub; Rytter, Wojciech; Waleń, Tomasz
Show all (5)
1 Citations
In the Manhattan Sequence Consensus problem (MSC problem) we are given k integer sequences, each of length ℓ, and we are to find an integer sequence x of length ℓ (called a consensus sequence), such that the maximum Manhattan distance of x from each of the input sequences is minimized. For binary sequences Manhattan distance coincides with Hamming distance, hence in this case the string consensus problem (also called string center problem or closest string problem) is a special case of MSC. Our main result is a practically efficient
$\mathcal{O}(\ell)$
time algorithm solving MSC for k ≤ 5 sequences. Practicality of our algorithms has been verified experimentally. It improves upon the quadratic algorithm by Amir et al. (SPIRE 2012) for string consensus problem for k = 5 binary strings. Similarly as in Amir’s algorithm we use a columnbased framework. We replace the implied general integer linear programming by its easy special cases, due to combinatorial properties of the MSC for k ≤ 5. We also show that for a general parameter k any instance can be reduced in linear time to a kernel of size k!, so the problem is fixedparameter tractable. Nevertheless, for k ≥ 4 this is still too much for any naive solution to be feasible in practice.
more …
By
Veloso, Manuela; Bowling, Michael; Achim, Sorin
1 Citations
One of the necessary steps in entering a smallsize RoboCup team is the actual construction of the robots. We have successfully built robots for RoboCup97 and RoboCup98, leading to two champion teams, namely CMUnited97 [2] and CMUnited98 [1].
By
Ponce, J.; Berg, T. L.; Everingham, M.; Forsyth, D. A.; Hebert, M.; Lazebnik, S.; Marszalek, M.; Schmid, C.; Russell, B. C.; Torralba, A.; Williams, C. K. I.; Zhang, J.; Zisserman, A.
Show all (13)
83 Citations
Appropriate datasets are required at all stages of object recognition research, including learning visual models of object and scene categories, detecting and localizing instances of these models in images, and evaluating the performance of recognition algorithms. Current datasets are lacking in several respects, and this paper discusses some of the lessons learned from existing efforts, as well as innovative ways to obtain very large and diverse annotated datasets. It also suggests a few criteria for gathering future datasets.
more …
By
Misra, Navodit; Blelloch, Guy; Ravi, R.; Schwartz, Russell
Show all (4)
Much modern work in phylogenetics depends on statistical sampling approaches to phylogeny construction to estimate probability distributions of possible trees for any given input data set. Our theoretical understanding of sampling approaches to phylogenetics remains far less developed than that for optimization approaches, however, particularly with regard to the number of sampling steps needed to produce accurate samples of tree partition functions. Despite the many advantages in principle of being able to sample trees from sophisticated probabilistic models, we have little theoretical basis for concluding that the prevailing sampling approaches do in fact yield accurate samples from those models within realistic numbers of steps. We propose a novel approach to phylogenetic sampling intended to be both efficient in practice and more amenable to theoretical analysis than the prevailing methods. The method depends on replacing the standard tree rearrangement moves with an alternative Markov model in which one solves a theoretically hard but practically tractable optimization problem on each step of sampling. The resulting method can be applied to a broad range of standard probability models, yielding practical algorithms for efficient sampling and rigorous proofs of accurate sampling for some important special cases. We demonstrate the efficiency and versatility of the method in an analysis of uncertainty in tree inference over varying input sizes. In addition to providing a new practical method for phylogenetic sampling, the technique is likely to prove applicable to many similar problems involving sampling over combinatorial objects weighted by a likelihood model.
more …
