The class of similarity based methods (SBM) covers most neural models and many other classifiers. Performance of such methods is significantly improved if irrelevant features are removed and feature weights introduced, scaling their influence on calculation of similarity. Several methods for feature selection and weighting are described. As an alternative to the global minimization procedures c...

Copyright c ©1997 IEEE. Reprinted from IEEE Information Theory Society Newsletter, Vol. 47, No. 1, March 1997. This material is posted here with permission of the IEEE. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obt...

We describe a sequential universal data compression procedure for binary tree sources that performs the \double mixture". Using a context tree, this method weights in an eecient recursive way the coding distributions corresponding to all bounded memory tree sources, and achieves a desirable coding distribution for tree sources with an unknown model and unknown parameters. Computational and stor...

................................................................................................................... i TABLE OF CONTENTS ..............................................................................................ii LIST OF TABLES .......................................................................................................iii LIST OF FIGURES .............................

zs denotes the concatenation of a data zz and a string s, l(s) is the length of a string s, 0 < ps < 1 is a constant given to a string s if 0 < l(s) 5 D 1, ,Bs = 0 if 2(s) = D (D > 1: some integer), Q,(x;) is the product of the predicted probabilities over the time instances such that the string defined below is s. The CTW assumes that the source is some tree source: the probability of x1+1 E A...

In this paper we introduce a so-called probability assignment technique for double exponential data based on weighting. Results for lossless compression of image data show that this improves the compression performance over methods based on probability distribution estimation. The weighting technique gives a robustness that aaects the compression favorably .

In [1], Alford et al. compared the performances of two Genetic and Evolutionary Methods (GEMs) for multibiometric feature selection and weighting. In this paper, we present two hybrid feature weighting/selection GEMs. Our results show that the hybrid GEMs outperform the GEMs presented in [1], using significantly fewer features while achieving practically the same recognition accuracy.

Abstract Propensity score weighting adjustment is commonly used to handle unit nonresponse. When the response mechanism is nonignorable in the sense that the response probability depends directly on the study variable, a followup sample is commonly used to obtain an unbiased estimator using the framework of two-phase sampling, where the follow-up sample is assumed to respond completely. In prac...

Automated spike detection methods for the epileptic EEG are highly desired to speed up and disambiguate EEG analysis. However, it is difficult to accurately and concisely present the performance of such algorithms due to the large number of recording and algorithm variables that must be accounted for. This paper summarizes the core variables involved and presents different methods for calculati...