Abstract — This paper presents a method for vision based estimation of the pose of human hands in interaction with objects. Despite the fact that most robotics applications of human hand tracking involve grasping and manipulation of objects, the majority of methods in the literature assume a free hand, isolated from the surrounding environment. Our hand tracking method is non-parametric, performing a nearest neighbor search in a large database (100000 entries) of hand poses with and without grasped objects. The system operates in real time, it is robust to self occlusions, object occlusions and segmentation errors, and provides full hand pose reconstruction from markerless video. Temporal consistency in hand pose is taken into account, without explicitly tracking the hand in the high dimensional pose space. I.

With improved sensors, the amount of data available in many vision problems has increased dramatically and allows the use of sophisticated learning algorithms to perform inference on the data. However, since these algorithms scale with data size, pruning the data is sometimes necessary. The pruning procedure must be statistically valid and a representative subset of the data must be selected without introducing selection bias. Information theoretic measures have been used for sampling the data, retaining its original information content. We propose an efficient Rényi entropy based subset selection algorithm. The algorithm is first validated and then applied to two sample applications where machine learning and data pruning are used. In the first application, Gaussian process regression is used to learn object pose. Here it is shown that the algorithm combined with the subset selection is significantly more efficient. In the second application, our subset selection approach is used to replace vector quantization in a standard object recognition algorithm, and improvements are shown. 1.

• A class of robust non-parametric learning methods • Involves a definition for a “kernel function ” [1] – Ex. Gaussian kernel: • Learning methods based on kernels involves – A weighted summation of kernel functions (K(x,x i)) – Solution of linear system based on kernel matrices • Scales O(N2) or O(N3) in time • O(N2) in memory • Objective: Use GPU to accelerate kernel machine learning approaches Proposed acceleration approach • Non-parametric way of estimating probability density function of a random variable [1] • Density, • Two popular kernels: Gaussian and Epanechnikov

Abstract. Kernel based approaches for machine learning have gained huge interest in the past decades because of their robustness. In some algorithms, the primary problem is the solution of a linear system involving the kernel matrix. Iterative Krylov approaches are often used to solve these efficiently [2, 3]. Fast matrix-vector products can be used to accelerate each Krylov iteration to further optimize the performance. In order to reduce the number of iterations of the Krylov approach, a preconditioner becomes necessary in many cases. Several researchers have proposed flexible preconditioning methods where the preconditioner changes with each iteration, and this class of preconditioners are shown to have good performance [6, 12]. In this paper, we use a Tikhonov regularized kernel matrix as a preconditioner for flexible GMRES [12] to solve kernel matrix based systems of equations. We use a truncated conjugate gradient (CG) method to solve the preconditioner system and further accelerate each CG iteration using fast matrix-vector products. The convergence of the proposed preconditioned GMRES is shown on synthetic data. The performance is further validated on problems in Gaussian process regression and radial basis function interpolation. Improvements are seen in each case.

Speaker recognition systems classify a test signal as a speaker or an imposter by evaluating a matching score between input and reference signals. We propose a new information theoretic approach for computation of the matching score using the Rényi entropy. The proposed entropic distance, the Kernelized Rényi distance (KRD), is formulated in a non-parametric way and the resulting measure is efficiently evaluated in a parallelized fashion on a graphical processor. The distance is then adapted as a scoring function and its performance compared with other popular scoring approaches in a speaker identification and speaker verification framework. Index Terms — Rényi entropy, similarity score, speaker recognition, GPU, fast algorithms 1.

Algorithms based on kernel methods play a central role in statistical machine learning. At their core are a number of linear algebra operations on matrices of kernel functions which take as arguments the training and testing data. These range from the simple matrix-vector product, to more complex matrix decompositions, and iterative formulations of these. Often the algorithms scale quadratically or cubically, both in memory and operational complexity, and as data sizes increase, kernel methods scale poorly. We use parallelized approaches on a multi-core graphical processor (GPU) to partially address this lack of scalability. GPUs are used to scale three different classes of problems, a simple kernelmatrix-vector product, iterative solution of linear systems of kernel function and QR and Cholesky decomposition of kernel matrices. Application of these accelerated approaches in scaling several kernel based learning approaches are shown, and in each case substantial speedups are obtained. The core software is released as an open source package, GPUML.

Abstract not found

Over the last decade, digital imaging has become ubiquitous. The advent of cheap digital cameras, and the inclusion of cameras in almost all mobile devices, has made photography one of the basic ways in which people record and communicate experiences. The ubiquity of cameras has imposed new constraints on their physical form. Camera modules are expected to be thin, light, and cheap. These restrictions make the production of high-quality images challenging. We turn to increasingly sophisticated algorithmic tools to transform the raw data captured by a camera into a photograph. This dissertation focuses on one such family of algorithmic tools: those expressible as a Gauss transform. One popular technique in this family is the bilateral filter, which smooths the fine detail in an image without crossing strong edges. It can be used to isolate and control the sharpness, tone, and contrast of a photograph at various scales. Its relatives, the joint-bilateral filter and the joint-bilateral upsample, allow for the fusion of data from multiple images. Another popular technique in the same family is non-local means, which denoises an image by replacing each pixel with the average color of all other pixels in the image with a similar local neighborhood.

people’s hands and feet using mixed