Perspective Taking as Egocentric Anchoring and Adjustment.

The authors propose that people adopt others' perspectives by serially adjusting from their own. As predicted, estimates of others' perceptions were consistent with one's own but differed in a manner consistent with serial adjustment (Study 1). Participants were slower to indicate that another's perception would be different from--rather than similar to--their own (Study 2). Egocentric biases increased under time pressure (Study 2) and decreased with accuracy incentives (Study 3). Egocentric biases also increased when participants were more inclined to accept plausible values encountered early in the adjustment process than when inclined to reject them (Study 4). Finally, adjustments tend to be insufficient, in part, because people stop adjusting once a plausible estimate is reached (Study 5). (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Functional theory of illusory conjunctions and neon colors.

Illusory conjunctions are the incorrect perceptual combination of briefly presented colors and shapes. In the neon colors illusion, achromatic figures take on the color of an overlaid grid of colored lines. Both illusions are explained by a theory that assumes (a) poor location information or poor spatial resolution for some aspects of visual information and (b) that the spatial location of features is constrained by perceptual organization. Computer simulations demonstrate that the mechanisms suggested by the theory are useful in veridical perception and they are sufficient to produce illusory conjunctions. The theory suggests mechanisms that economically encode visual information in a way that filters noise and fills in missing data. Issues related to neural implementation are discussed. Four experiments illustrate the theory. Illusory conjunctions are shown to be affected by objective stimulus organization, by subjective organization, and by the linguistic structure of ambiguous Hebrew words. Neon colors are constrained by linguistic structure in the same way as illusory conjunctions. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

General Perfectly Periodic Scheduling

In a perfectly periodic schedule, each job must be scheduled precisely every some fixed number of time units after its previous occurrence. Traditionally, motivated by centralized systems, the perfect periodicity requirement is relaxed, the main goal being to attain the requested average rate. Recently, motivated by mobile clients with limited power supply, perfect periodicity seems to be an attractive alternative that allows clients to save energy by reducing their "busy waiting" time. In this case, clients may be willing to compromise their requested service rate in order to get perfect periodicity. In this paper we study a general model of perfectly periodic schedules, where each job has a requested period and a length; we assume that m jobs can be served in parallel for some given m. Job lengths may not be truncated, but granted periods may be different than the requested periods. We present an algorithm which computes schedules such that the worst-case proportion between the requested period and the granted period is guaranteed to be close to the lower bound. This algorithm improves on previous algorithms for perfect schedules in providing a worst-case guarantee rather than an average-case guarantee, in generalizing unit length jobs to arbitrary length jobs, and in generalizing the single-server model to multiple servers.     

The value of agreement a new boosting algorithm

In the past few years unlabeled examples and their potential advantage have received a lot of attention. In this paper a new boosting algorithm is presented where unlabeled examples are used to enforce agreement between several different learning algorithms. Not only do the learning algorithms learn from the given training set but they are supposed to do so while agreeing on the unlabeled examples. Similar ideas have been proposed before (for example, the Co-Training algorithm by Mitchell and Blum), but without a proof or under strong assumptions. In our setting, it is only assumed that all learning algorithms are equally adequate for the tasks. A new generalization bound is presented where the use of unlabeled examples results in a better ratio between training-set size and the resulting classifier's quality and thus reduce the number of labeled examples necessary for achieving it. The extent of this improvement depends on the diversity of the learners—a more diverse group of learners will result in a larger improvement whereas using two copies of a single algorithm gives no advantage at all. As a proof of concept, the algorithm, named Agreement Boost, is applied to two test problems. In both cases, using Agreement Boost results in an up to 40% reduction in the number of labeled examples.     

Approximate distributed top-<Emphasis Type="Italic">k</Emphasis> queries

We consider a distributed system where each node keeps a local count for items (similar to elections where nodes are ballot boxes and items are candidates). A top-k query in such a system asks which are the k items whose global count, across all nodes in the system, is the largest. In this paper, we present a Monte Carlo algorithm that outputs, with high probability, a set of k candidates which approximates the top-k items. The algorithm is motivated by sensor networks in that it focuses on reducing the individual communication complexity. In contrast to previous algorithms, the communication complexity depends only on the global scores and not on the partition of scores among nodes. If the number of nodes is large, our algorithm dramatically reduces the communication complexity when compared with deterministic algorithms. We show that the complexity of our algorithm is close to a lower bound on the cell-probe complexity of any non-interactive top-k approximation algorithm. We show that for some natural global distributions (such as the Geometric or Zipf distributions), our algorithm needs only polylogarithmic number of communication bits per node. An extended abstract of this paper appeared in Proc. 13th Int. Colloquium on Structural Information and Communication Complexity, SIROCCO 2006, Lecture Notes in Computer Science 4056, pp. 319–333.     

Creating a consensus ranking of proposals from reviewers’ partial ordinal rankings

Peer review of research proposals and articles is an essential element in R&D processes worldwide. In most cases, each reviewer evaluates a small subset of the candidate proposals. The review board is then faced with the challenge of creating an overall “consensus” ranking on the basis of many partial rankings. In this paper we propose a branch-and-bound model to support the construction of an aggregate ranking from the partial rankings provided by the reviewers. In a recent paper we proposed ways to allocate proposals to reviewers so as to achieve the maximum possible overlap among the subsets of proposals allocated to different reviewers. Here, we develop a special branch-and-bound algorithm that utilizes the overlap generated through our earlier methods to enable discrimination in ranking the competing proposals. The effectiveness and efficiency of the algorithm is demonstrated with small numerical examples and tested through an extensive simulation experiment.     

Video distribution under multiple constraints

We consider the optimization problem of providing a set of video streams to a set of clients, where each stream has costs in m possible measures (such as communication bandwidth, processing bandwidth, etc.), and each client has its own utility function for each stream. We assume that the server has a budget cap on each of the m cost measures; each client has an upper bound on the utility that can be derived from it, and potentially also upper bounds in each of the m cost measures. The task is to choose which streams the server will provide, and out of this set, which streams each client will receive. The goal is to maximize the overall utility subject to the budget constraints. We give an efficient approximation algorithm with approximation factor of O(m) with respect to the optimal possible utility for any input, assuming that clients have only a bound on their maximal utility. If, in addition, each client has at most m_c capacity constraints, then the approximation factor increases by another factor of O(m_clogn), where n is the input length. We also consider the special case of “small” streams, namely where each stream has cost of at most O(1/logn) fraction of the budget cap, in each measure. For this case we present an algorithm whose approximation ratio is O(logn).     

Knowledge-based part correspondence

This paper presents a direct method for finding corresponding pairs of parts between two shapes. Statistical knowledge about a large number of parts from many different objects is used to find a part correspondence between two previously unseen input shapes. No class membership information is required. The knowledge-based approach is shown to produce significantly better results than a classical metric distance approach. The potential role of part correspondence as a complement to geometric and structural comparisons is discussed.     

Spatial Aliasing in Spherical Microphone Arrays

Performance of microphone arrays at the high-frequency range is typically limited by aliasing, which is a result of the spatial sampling process. This paper presents analysis of aliasing for spherical microphone arrays, which have been recently studied for a range of applications. The paper presents theoretical analysis of spatial aliasing for various sphere sampling configurations, showing how high-order spherical harmonic coefficients are aliased into the lower orders. Spatial antialiasing filters on the sphere are then introduced, and the performance of spatially constrained filters is compared to that of the ideal antialiasing filter. A simulation example shows how the effect of aliasing on the beam pattern can be reduced by the use of the antialiasing filters