Using a similarity measure for credible classification

Abstract

This paper concerns classification by Boolean functions. We investigate the classification accuracy obtained by standard classification techniques on unseen points (elements of the domain, {0, 1}n, for some n) that are similar, in particular senses, to the points that have been observed as training observations. Explicitly, we use a new measure of how similar a point x ∈ {0, 1}n is to a set of such points to restrict the domain of points on which we offer a classification. For points sufficiently dissimilar, no classification is given. We report on experimental results which indicate that the classification accuracies obtained on the resulting restricted domains are better than those obtained without restriction. These experiments involve a number of standard data-sets and classification techniques. We also compare the classification accuracies with those obtained by restricting the domain on which classification is given by using the Hamming distance.

A hybrid digital image watermarking algorithm based on DCT‐DWT and auto‐thresholding

“…Peak signal‐to‐noise ratio minimum gained value was 42.74 dB in the peppers image, and the maximum value was 43.46 dB in the baboon image. This result indicate high visual quality and high imperceptible algorithm, because a PSNR of 30 and above is considered the acceptable value for image quality evaluation .…”

Rank thresholds in classifier ensembles in medical diagnosis

“…These models are often extensions of previously known linear methods, allowing them to operate on non-linear feature spaces. Some examples of non-linear classifiers are Multi-Layer Neural Networks [12], Bayesian Networks [12,13] or similarity-based methods [14].…”

A new imputation method for incomplete binary data

“…The technique we introduce employs a ''similarity measure'' introduced in [1]. The Boolean similarity measure has already proven to be of some application in classification problems [19]. Here, we use it to help indicate whether a missing value should be 0 or 1, and we compare experimentally the performance of our technique with ones based on the usual Hamming distance measure and MI technique using SAS [17].…”