outperform existing feature sets for human detection. We study the influence of each stage of the computation on performance, concluding that fine-scale gradients, fine orientation binning, relatively coarse spatial binning, and high-quality local contrast normalization in overlapping descriptor blocks

the picture plane. Surface shading calculations are performed at every voxel with local gradient vec-tors serving as surface normals. In a separate step, surface classification operators are applied to obtain a partial opacity for every voxel. Operators that detect isovalue contour surfaces and region

. Thus, we examine how changes in credit spreads respond to proxies for both changes in the probability of future default and for changes in the recovery rate. Separately, recent empirical studies find that the corporate bond market tends to have relatively high transactions costs and low volume. 1

and potentially further optimized through inline-expansion. By examining the complete inheritance graph of a program, which we call class hierarchy analysis, the compiler can improve the quality of static class information and thereby improve run-time performance. In this paper we present class hierarchy analysis

, and also corrects for the correlations between neighboring voxels. This makes the same approach suitable for the spatial normalization of both high-quality magnetic resonance images, and low-resolution noisy positron emission tomography images. A fast algorithm has been developed that utilizes Taylor’s

We describe a mechanism for scalable control of multicast continuous media streams. The mechanism uses a novel probing mechanism to solicit feedback information in a scalable manner and to estimate the number of receivers. In addition, it separates the congestion signal from the congestion control

with phylogenetic methods, but these are slow and dif®cult to automate. Automatic clustering methods based on two-way best genome-wide matches on the other hand, have so far not separated in-paralogs from out-paralogs effectively. We present a fully automatic method for ®nding orthologs and inparalogs from two

show how technology can be seen in terms of experience with technological artifacts. This approach orients us toward the felt-life of technology-toward engagement, enchantment, irritation, and fulfillment. But we also recognize that the feeling-life does not begin and end with the immediate quality

..................................................................... ....... i Acknowledgements ............................................................... ii 1. Introduction .................................................................... 1 2. Quality of Service Terminology, Principles

with independent subspace analysis and basic nonnegative matrix factorization, which are based on the same linear model. According to these simulations, the proposed method enables a better separation quality than the previous algorithms. Especially, the temporal continuity criterion improved the detection