The emergence of the world-wide-web has led to an increased interest in methods for searching for information. A key characteristic of many of the online document collections is that the documents have predefined category information, for example, the variety of scientific articles accessible via digital libraries (e.g., ACM, IEEE, etc.), medical articles, news-wires, and various directories (e...

In this paper we revisit the practice of using feature selection for dimensionality and noise reduction. Commonly we score features according to some weighting scheme and then specify that the top N ranked features or top N percents of scored features are to be used for further processing. In text classification, such a selection criteria lead to significantly different sizes of (unique) featur...

In the context of supervised learning a function by neural network, we claim and empirically verify that network yields better results when distribution data set focuses on regions where to learn is steep. We first traduce this assumption in mathematically workable way using Taylor expansion emphasize new training based derivatives learn. Then, theoretical derivations allow construction methodo...

The present paper investigates the choice of objective weighting schemes for computing the Purchasing Managers’ Index (PMI). To this end, the principal components variable selection strategy considered by Jolliffee (1972, 1973) is applied to monthly data of five key diffusion indexes for the period from January 1948 to October 2004, compiled by the Institute for Supply Management (ISM). It turn...

In this work we demonstrate an improvement in the state-of-theart large vocabulary continuous speech recognition (LVCSR) performance, under clean and noisy conditions, by the use of visual information, in addition to the traditional audio one. We take a decision fusion approach for the audio-visual information, where the single-modality (audioand visualonly) HMM classifiers are combined to reco...

We propose a new theory that quantifies information in probability distributions and derive a new document representation model for text clustering. By extending Shannon entropy to accommodate a non-linear relation between information and uncertainty, the proposed Least Information theory (LIT) provides insight into how terms can be weighted based on their probability distributions in documents...

Most sentiment analysis approaches use as baseline a support vector machines (SVM) classifier with binary unigram weights. In this paper, we explore whether more sophisticated feature weighting schemes from Information Retrieval can enhance classification accuracy. We show that variants of the classic tf.idf scheme adapted to sentiment analysis provide significant increases in accuracy, especia...

Pharmacokinetic parameters are usually determined by noncompartmental methods or by assuming a multicompartment model in order to describe the behaviour of the drug in the patient's body. With a noncompartmental analysis the terminal half-life and the volume of distribution are derived from log-linear regression of the terminal elimination phase, whereas the estimation of these parameters is ba...