We present a general methodology for indexing with multivariate features based on the Bhattacharyya distance, which takes into account both the mean-di erence and the covariancedi erence between two distributions. To reduce the amount of computations and the size of logical database entry, we approximate the Bhattacharyya distance in the low dimensional subspace of the rst few principal compone...

The risk di$erence is frequently used as a measure of the actual gain in the success rate between two treatments within a center (i.e. hospital). Interest is devoted to combining the risk di$erence across several centers under homogeneity but allowing for baseline-risk heterogeneity in each of the treatment arms. The purpose is to compare the e5ciency of six estimators for the common risk di$er...

This paper describes the impact of traÆc handling mechanisms on network capacity for support of Quality of Service (QoS) in the Internet. The emergence of applications with diverse throughput, loss and delay requirements requires a network that is capable of supporting di erent levels of service as opposed to the single best-e ort service that was the foundation of the Internet. As a result the...

A critical problem in decision analysis is how to evaluate the di€erence between two or more di€erent rankings of a set of alternatives. In this paper it is assumed that a group of decision makers has already established a ranking for each one of a set of alternatives. Then, the problem examined here is how to evaluate the di€erences among these rankings. This problem does not have a unique sol...

This paper investigates several di erent methods for performing vocal tract length normalisation (VTLN) which are either completely linear or piece-wise linear. Furthermore the combination of VTLN with either standard unconstrained maximum likelihood linear regression (MLLR) or constrained MLLR is considered. Results on the Switchboard corpus show that there is little di erence in performance b...

The impact of traÆc handling mechanisms on network capacity and supporting of Quality of Service (QoS) in the Internet is studied. The emergence of applications with diverse throughput, loss and delay requirements requires a network that is capable of supporting di erent levels of service as opposed to the single best-e ort service that was the foundation of the Internet. As a result the Integr...

We describe the two most commonly used systems for induction of decision trees for classi cation: C4.5 and CART. We highlight the methods and di erent decisions made in each system with respect to splitting criteria, pruning, noise handling, and other di erentiating features. We describe how rules can be derived from decision trees and point to some di erence in the induction of regression tree...

The equivalent, second equivalent and (simply) modi®ed equations for the implicit midpoint rule are shown to be asymptotically equivalent in the sense that an asymptotic analysis of these equations with the time step size as small parameter yields exactly the same results; for linear problems with constant coecients, they are also equivalent to the original ®nite di€erence scheme. Straightforw...

Prediction and ltering of continuous-time stochastic processes require a solver of a continuous-time di erential Lyapunov equation (cdle). Even though this can be recast into an ordinary di erential equation (ode), where standard solvers can be applied, the dominating approach in Kalman lter applications is to discretize the system and then apply the discrete-time di erence Lyapunov equation (d...

In this paper we present an analysis of three di erent formulations of the discontinuous Galerkin method for di usion equations. The rst formulation yields an inconsistent and weakly unstable scheme, while the other two formulations, the local discontinuous Galerkin approach and the Baumann-Oden approach, give stable and convergent results. When written as nite di erence schemes, such a distinc...