Indoor location systems based on IEEE 802.11b (WiFi) mobile devices often rely the received signal strength indicator to estimate user position. Two key characteristics of these have not yet been fully analyzed, namely, temporal and spatial sampling process required adequately describe distribution electromagnetic field in indoor scenarios; device calibration, necessary for supporting different within same system. By using a previously proposed nonparametric methodology system comparison, we...

Unpredictable topology changes, energy constraints and link unreliability make the information transmission a challenging problem in wireless sensor networks (WSN). Taking some ideas from machine learning methods, we propose novel geographic routing algorithm for WSN, named Q-probabilistic (Q-PR), that makes intelligent decisions delayed reward of previous actions local interaction among neighbor nodes, by using reinforcement Bayesian decision model. Moreover, considering message importance...

Indoor location (IL) using received signal strength (RSS) is receiving much attention, mainly due to its ease of use in deployed IEEE 802.11b (Wi-Fi) wireless networks. Fingerprinting the most widely used technique. It consists estimating position by comparison a set RSS measurements, made mobile device, with database measurements whose locations are known. However, convenient data structure be and actual performance proposed fingerprinting algorithms still controversial. In addition,...

Communication skills development is one of the main goals engineering education. We propose an integrated student-centered collaborative learning environment for developing communication skills, using project-based methods and peer assessment. In our environment, projects are assigned to small groups students under teacher supervision, documented in a wiki-editing tool presented during public poster session. By combining wiki entries presentations, we intend facilitate (1) gain access...

Decision theory shows that the optimal decision is a function of posterior class probabilities. More specifically, in binary classification, based on comparison probabilities with some threshold. Therefore, most accurate estimates are required near these thresholds. This paper discusses design objective functions provide more probability values, taking into account characteristics each problem. We propose learning algorithms stochastic gradient minimization loss functions. show performance...

High-Power electric grid networks require extreme security in their associated telecommunication network to ensure protection and control throughout power transmission. Accordingly, supervisory data acquisition systems form a vital part of any critical infrastructure, the safety from intrusion is crucial. Whereas events related operation maintenance are often available carefully documented, only some tools have been proposed discriminate information dealing with heterogeneous detection...

Human skin detection in color images is a key preprocessing stage many image processing applications. Though kernel-based methods have been recently pointed out as advantageous for this setting, there still few evidence on their actual superiority. Specifically, binary Support Vector Classifier (two-class SVM) and one-class Novelty Detection (SVND) only tested some example or limited databases. We hypothesize that comparative performance evaluation representative application-oriented...

Abstract Many types of nonlinear classifiers have been proposed to automatically generate land-cover maps from satellite images. Some are based on the estimation posterior class probabilities, whereas others estimate decision boundary directly. In this paper, we propose a modular design able focus learning process by using probability estimates. To do so, use self-configuring architecture that incorporates specialized modules deal with conflicting classes, and apply algorithm focuses regions...

Analyzing the structure of family cost functions that are minimum when classifier outputs equal to class probabilities, we found all them can be expressed as sum a generalized entropy measure and an error component. This suggests novel algorithm for classification uses both labeled unlabeled data is based on following idea: use minimize function corresponding measure. minimization principle applied terrain Landsat images.

In some real applications, such as medical diagnosis or remote sensing, available training data do not often reflect the true a priori probabilities of underlying distribution. The classifier designed from these may be suboptimal. Building classifiers that are robust against changes in prior is possible by applying minimax learning strategy. this paper, we propose simple fixed-point algorithm able to train neural [i.e., minimizing worst (maximum) risk]. Moreover, present new parametric...

Satellite-derived soil moisture (SM) products have become an important information source for the study of land surface processes in hydrology and monitoring. Characterizing estimating memory persistence from satellite observations is paramount relevance, has deep implications ecology, water management, climate modeling. In this work, we address problem SM estimation microwave sensors using several autocorrelation metrics that, unlike traditional approaches, build on accurate estimates...

Doppler ultrasound M-mode images are routinely used in clinical echocardiography, and they have been proposed for non-invasive estimation of the intracardiac pressure gradients heart, a process that has shown to be sensitive spline interpolation.In this work, we scrutinized effect interpolation with new approach using support vector machines (SVM) autocorrelation Mercer kernels images.The SVM algorithm was modified provide whole image terms reduced set pixels as training data set.The color...

A new method for non-uniform interpolation of electroanatomical cardiac maps from Cardiac Navigation Systems (CNS) is here proposed and benchmarked.We adapted the equations support vector machines estimation problems in terms two angular dimensions azimuth elevation used an autocorrelation kernel.Moreover, influence number spatial locations, its minimum to obtain a map that precisely replicates original or gold-standard effect working 2D 3D were also studied.Two basic simulation scenarios...

This paper addresses the problem of classifying multispectral images when a priori knowledge about classes is not complete: true number known, or it possible to obtain ground truth data for some in image. We propose method perform image classification taking into account all classes, "known" and "unknown", based on accurate estimates prior probabilities joint probability density functions (pdfs). To this end, we application dependence tree approximation mitigate few available samples....

Ventricular fibrillation (VF) signals are characterized by highly volatile and erratic electrical impulses, the analysis of which is difficult given complex behavior heart rhythms in left (LV) right ventricles (RV), as sometimes shown intracardiac recorded Electrograms (EGM). However, there few studies that analyze VF humans according to simultaneous two ventricles. The objective this work was perform a spectral non-linear recordings 22 patients with Congestive Heart Failure (CHF) clinical...

The problem of identifying terrains in Landsat-TM images on the basis non-uniformly distributed labeled data is discussed this paper. Our approach based use neural network classifiers that learn to predict posterior class probabilities. Principal Component Analysis (PCA) used extract features from spectral and contextual information. proposed scheme obtains lower error rates other model-based approaches.