We introduce new transforms for efficient compression of image blocks with directional preferences. Each transform, which is an orthogonal basis for a specific direction, is constructed from an eigen-decomposition of a discrete directional Laplacian system matrix. The method is a natural extension of the DCT, expressing the Laplacian in Cartesian coordinates rotated to some predetermined angles. Symmetry properties of the transforms over square domains lead to efficient computation and compact storage of the directional transforms. A version of the directional transforms was implemented within the beyond HEVC software and demonstrated significant improvement for intra block coding.

We present a new out-of-core sparse symmetric-indefinite factorization algorithm. The most significant innovation of the new algorithm is a dynamic partitioning method for the sparse factor. This partitioning method results in very low I/O traffic and allows the algorithm to run at high computational rates, even though the factor is stored on a slow disk. Our implementation of the new code compares well with both high-performance in-core sparse symmetric-indefinite codes and a high-performance out-of-core sparse Cholesky code.

In this paper we present a speaker recognition algorithm that models explicitly intra-speaker inter-session variability. Such variability may be caused by changing speaker characteristics (mood, fatigue, etc.), channel variability or noise variability. We define a session-space in which each session (either train or test session) is a vector. We then calculate a rotation of the session-space for which the estimated intra-speaker subspace is isolated and can be modeled explicitly. We evaluated our technique on the NIST-2004 speaker recognition evaluation corpus, and compared it to a GMM baseline system. Results indicate significant reduction in error rate. 1.