The bottle-neck in most cloth simulation systems is that time steps must be small to avoid numerical instability. This paper describes a cloth simulation system that can stably take large time steps. The simulation system couples a new technique for enforcing constraints on individual cloth particles with an implicit integration method. The simulator models cloth as a triangular mesh, with internal cloth forces derived using a simple continuum formulation that supports modeling operations such as local anisotropic stretch or compression; a unified treatment of damping forces is included as well. The implicit integration method generates a large, unbanded sparse linear system at each time step which is solved using a modified conjugate gradient method that simultaneously enforces particles' constraints. The constraints are always maintained exactly, independent of the number of conjugate gradient iterations, which is typically small. The resulting simulation system is significantly fast...

Conditional random fields for sequence labeling offer advantages over both generative models like HMMs and classifiers applied at each sequence position. Among sequence labeling tasks in language processing, shallow parsing has received much attention, with the development of standard evaluation datasets and extensive comparison among methods. We show here how to train a conditional random field to achieve performance as good as any reported base noun-phrase chunking method on the CoNLL task, and better than any reported single model. Improved training methods based on modern optimization algorithms were critical in achieving these results. We present extensive comparisons between models and training methods that confirm and strengthen previous results on shallow parsing and training methods for maximum-entropy models.

Maximum Margin Matrix Factorization (MMMF) was recently suggested (Srebro et al., 2005) as a convex, infinite dimensional alternative to low-rank approximations and standard factor models. MMMF can be formulated as a semi-definite programming (SDP) and learned using standard SDP solvers. However, current SDP solvers can only handle MMMF problems on matrices of dimensionality up to a few hundred. Here, we investigate a direct gradient-based optimization method for MMMF and demonstrate it on large collaborative prediction problems. We compare against results obtained by Marlin (2004) and find that MMMF substantially outperforms all nine methods he tested. 1.

We have been collecting and recording routing paths from a test host to each of over 90,000 registered networks on the Internet since August 1998. The resulting database contains interesting routing and reachability information, and is available to the public for research purposes. The daily scans cover approximately a tenth of the networks on the Internet, with a full scan run roughly once a month. We have also been collecting Lucent's intranet data, and applied these tools to understanding its size and connectivity. We have also detecting the loss of power to routers in Yugoslavia as the result of NATO bombing. A simulated spring-force algorithm lays out the graphs that results from these databases. This algorithm is well known, but has never been applied to such a large problem. The Internet graph, with around 88,000 nodes and 100,000 edges, is much larger than those previously considered tractable by the data visualization community. The resulting Internet layouts are pleasant, though rather cluttered. On smaller networks, like Lucent's intranet, the layouts present the data in a useful way. For the Internet data, we have tried plotting a minimum distance spanning tree; by throwing away edges, the remaining graph can be made more accessible. Once a layout is chosen, it can be colored in various ways to show network-relevant data, such as IP address, domain information, location, ISPs, location of firewalls, etc. This paper expands and updates the description of the project given in [2]. 1

2 Everything Old Is New Again: A Fresh Look at Historical

The most recent revision of this document will always be found at

Abstract — This paper describes a vision-based, large-area, simultaneous localization and mapping (SLAM) algorithm that respects the low-overlap imagery constraints typical of underwater vehicles while exploiting the inertial sensor information that is routinely available on such platforms. We present a novel strategy for efficiently accessing and maintaining consistent covariance bounds within a SLAM information filter, thereby greatly increasing the reliability of data association. The technique is based upon solving a sparse system of linear equations coupled with the application of constant-time Kalman updates. The method is shown to produce consistent covariance estimates suitable for robot planning and data association. Real-world results are presented for a vision-based 6-DOF SLAM implementation using data from a recent ROV survey of the wreck of the RMS Titanic. I.

Abstract — This paper presents the novel insight that the SLAM information matrix is exactly sparse in a delayedstate framework. Such a framework is used in view-based representations of the environment which rely upon scanmatching raw sensor data. Scan-matching raw data results in virtual observations of robot motion with respect to a place its previously been. The exact sparseness of the delayed-state information matrix is in contrast to other recent featurebased SLAM information algorithms like Sparse Extended Information Filters or Thin Junction Tree Filters. These methods have to make approximations in order to force the feature-based SLAM information matrix to be sparse. The benefit of the exact sparseness of the delayed-state framework is that it allows one to take advantage of the information space parameterization without having to make any approximations. Therefore, it can produce equivalent results to the “full-covariance ” solution. Index Terms — Delayed states, EIF, SLAM. I.

Most literature on Support Vector Machines (SVMs) concentrate on the dual optimization problem. In this paper, we would like to point out that the primal problem can also be solved efficiently, both for linear and non-linear SVMs, and that there is no reason for ignoring this possibilty. On the contrary, from the primal point of view new families of algorithms for large scale SVM training can be investigated.

Abstract not found