Emilio Soria

Fetal ECG extraction using an FIR neural network

Non-invasive electrocardiography reveals itself as a very interesting method to obtain reliable information about the state of the fetus, thus assuring its well-being during pregnancy. In this paper, a finite impulse response (FIR) neural network is included in the familiar adaptive noise cancellation scheme in order to provide highly nonlinear dynamic capabilities to the recovery model. A novel methodology for selecting the optimal topology is also presented. Results from its application to both simulated and real registers are shown and benchmarked with the classical LMS (least mean squares) and normalized LMS (NLMS) algorithms. Outcomes indicate that the FIR network is a reliable method for the fetal electrocardiogram recovery.

Predicting service request in support centers based on nonlinear dynamics, ARMA modeling and neural networks

In this paper, we present the use of different mathematical models to forecast service requests in support centers (SCs). A successful prediction of service request can help in the efficient management of both human and technological resources that are used to solve these eventualities. A nonlinear analysis of the time series indicates the convenience of nonlinear modeling. Neural models based on the time delay neural network (TDNN) are benchmarked with classical models, such as auto-regressive moving average (ARMA) models. Models achieved high values for the correlation coefficient between the desired signal and that predicted by the models (values between 0.88 and 0.97 were obtained in the out-of-sample set). Results show the suitability of these approaches for the management of SCs.

Neural networks for animal science applications: Two case studies

Abstract Artificial neural networks have shown to be a powerful tool for system modelling in a wide range of applications. In this paper, we focus on neural network applications to intelligent data analysis in the field of animal science. Two classical applications of neural networks are proposed: time series prediction and clustering. The first task is related to the prediction of weekly milk production in goat flocks, which includes a knowledge discovery stage in order to analyse the relative relevance of the different variables. The second task is the clustering of goat flocks; it is used to analyse different livestock surveys by using self-organizing maps and the adaptive resonance theory, thus obtaining a qualitative knowledge from these surveys. Achieved results show the usefulness of neural networks in two animal science applications.

Feature selection of hyperspectral data through local correlation and SFFS for crop classification

In this paper, we propose a procedure to reduce dimensionality of hyperspectral data while preserving relevant information for posterior crop cover classification. One of the main problems with hyperspectral image processing is the huge amount of data involved. In addition, pattern recognition methods are sensitive to problems associated to high dimensionality feature spaces (referred to as Hughes phenomenon of curse of dimensionality). We propose a dimensionality reduction strategy that eliminates redundant information by means of local correlation criterion between contiguous spectral bands; and a subsequent selection of the most discriminative features based on a Sequential Float Feature Selection algorithm. This method is tested with a crop cover recognition application of six hyperspectral images from the same area acquired with the 128-bands HyMap spectrometer during the DAISEX99 campaign. In the experiments, we analyze the dependence on the dimension and employed metrics. The results obtained using the Gaussian Maximum Likelihood improve the classification accuracy and confirm the validity of the proposed approach. Finally, we analyze the selected bands of the input space on order to gain knowledge on the problem and to give a physical interpretation of the results.

CART-based feature selection of hyperspectral images for crop cover classification

In this paper, we propose a procedure to reduce data dimensionality while preserving relevant information for posterior crop cover classification. The huge amount of data involved in hyperspectral image processing is one of the main problems in order to apply pattern recognition techniques. We propose a dimensionality reduction strategy that eliminates redundant information and a subsequent selection of the most discriminative features based on classification and regression trees (CART). CART allow feature selection based on the classification success, it is a non-linear method and specially allows knowledge discovery. The main advantage of our proposal relies on model interpretability, since we can get qualitative information by analyzing the surrogate and main splits of the tree. This method is tested with a crop cover recognition application of six hyperspectral images from the same area acquired with the 128-bands HyMap spectrometer. Even though CART do not provide the best results in classification it is useful for a previous pre-processing step of feature selection. Finally, we analyze the selected bands of the input space in order to gain knowledge on the problem and to give a physical interpretation of results.

ETAT: Expository Text Analysis Tool

Qualitative methods that analyze the coherence of expository texts not only are time consuming, but also present challenges in collecting data on coding reliability. We describe software that analyzes expository texts more rapidly and produces a notable level of objectivity ETAT (Expository Text Analysis Tool) analyzes the coherence of expository texts. ETAT adopts a symbolic representational system, known asconceptual graph structures. ETAT follows three steps: segmentation of a text into nodes, classification of the unidentified nodes, and linking the nodes with relational arcs. ETAT automatically constructs a graph in the form of nodes and their interrelationships, along with various attendant statistics and information about noninterrelated, isolated nodes. ETAT was developed in Java, so it is compatible with virtually all computer systems.

Opposite effects of myocardial stretch and verapamil on the complexity of the ventricular fibrillatory pattern: an experimental study.

CHORRO, F.J., et al.: Opposite Effects of Myocardial Stretch And Verapamil on The Complexity of The Ventricular Fibrillatory Pattern: An Experimental Study. An experimental model is used to analyze the effects of ventricular stretching and verapamil on the activation patterns during VF. Ten Langendorff‐perfused rabbit hearts were used to record VF activity with an epicardial multiple electrode before, during, and after stretching with an intraventricular balloon, under both control conditions and during verapamil (Vp) infusion (0.4–0.8 μmol). The analyzed parameters were dominant frequency (FrD) spectral analysis, the median (MN) of the VF intervals, and the type of activation maps during VF (I = one wavelet without block lines, II = two simultaneous wavelets with block lines, III = three or more wavelets with block lines). Stretch accelerates VF (FrD: 22.8 ± 6.4 vs 15.2 ± 1.0 Hz, P < 0.01; MN: 48 ± 13 vs 68 ± 6 ms, P < 0.01). On fitting the FrD time changes to an exponential model after applying and suppressing stretch, the time constants (stretch: 101.2 ± 19.6 s; stretch suppression: 97.8 ± 33.2 s) do not differ significantly. Stretching induces a significant variation in the complexity of the VF activation maps with type III increments and type I and II decrements (control: I = 17.5%, II = 50.5%, III = 32%; stretch: I = 7%, II = 36.5%, III = 56.5%, P < 0.001). Vp accelerates VF (FrD: 20.9 ± 1.9 Hz, P < 0.001 vs control; MN: 50 ± 5 ms, P < 0.001 vs control) and diminishes activation maps complexity (I = 25.5%, II = 60.5%, III = 14%, P < 0.001 vs control). On applying stretch during Vp perfusion, the fibrillatory process is not accelerated to any greater degree. However, type I and II map decrements and type III increments are recorded, though reaching percentages similar to control (I = 16.5%, II = 53%, III = 30.5%, NS vs control). The following conclusions were found: (1) myocardial stretching accelerates VF and increases the complexity of the VF activation pattern; (2) time changes in the FrD of VF during and upon suppressing stretch fit an exponential model with similar time constants; and (3) although stretching and verapamil accelerate the VF process, they exert opposite effects upon the complexity of the fibrillatory pattern.

A novel approach to introducing adaptive filters based on the LMS algorithm and its variants

This paper presents a new approach to introducing adaptive filters based on the least-mean-square (LMS) algorithm and its variants in an undergraduate course on digital signal processing. Unlike other filters currently taught to undergraduate students, these filters are nonlinear and time variant. This proposal introduces adaptive filtering in the context of a linear time-invariant system using a real problem. In this way, introducing adaptive filters using concepts already familiar to the students motivates their interest through practical application. The key point for this simplification is that the input to the filter is constant so that the adaptive filter becomes linear. Therefore, a complete arsenal of mathematical tools, already known by the students, is available to analyze the performance of the filters and obtain the key parameters to adaptive filters, e.g., speed of convergence and stability. Several variants of the basic LMS algorithm are described the same way.

BioLab: An Educational Tool for Signal Processing Training in Biomedical Engineering

This paper introduces and evaluates BioLab, a tool for teaching biosignal processing. BioLab has been used in the biomedical engineering module that is given in the second semester of the fifth year of the electronic engineering degree at the University of Valencia, Spain. This module and its correspondent curricular pathway are also reviewed. BioLab allows the results obtained with digital processing techniques to be shown interactively in the theory classes, and it also provides support in laboratory sessions. The graphic interface of BioLab simplifies its learning and use and provides access to processing and visualization functions by means of menus. The tool is based on Matlab since the students have had previous experience in this environment. BioLab also supports diverse formats of data files, which facilitate access to real records and their conversion to usable formats. The modular structure of BioLab allows it to be easily extended to other educational materials that are related to signal processing and to research applications. An evaluation of BioLab has revealed that students found it useful for understanding the general concepts of digital processing and biosignal processing in particular. The students also found BioLab very easy to learn and use

Feature selection using ROC curves on classification problems

Feature Selection (FS) is one of the key stages in classification problems. This paper proposes the use of the area under Receiver Operator Characteristic curves to measure the individual importance of every input as well as a method to discover the variables that yield a statistically significant improvement in the discrimination power of the classification model.