Cristian Guarnizo

Optimal sampling frequency in wavelet-based signal feature extraction using particle swarm optimization

A methodology for optimum sampling frequency selection for wavelet feature extraction is presented. We show that classification accuracy is enhanced by adequately selecting the parameters: number of decomposition levels, wavelet function and sampling rate. A novel approach for selecting the parameters based on particle swarm optimization (PSO) is presented. Experimental results conducted on two different datasets with support vector machine (SVM) classifiers confirm the superiority and advantages of the proposed method. It is shown empirically that the proposed method outperforms significantly the existing methods in terms of accuracy rate.

Microelectrode Signals Segmentation Using Stationary Wavelet Transform

We present a methodology for the automatic segmentation of extracellular microelectrode recordings (MER) based on stationary wavelet transform and modified F test which identify segments with equal time - frequency behavior. The method was tested using synthetic signals and then applied to real MER signals, achieving artifact removal and showing a superior performance than segmentation based on time representation.

ANÁLISIS DE REDUCCIÓN DE RUIDO EN SEÑALES PROVENIENTES DE MICROELECTRODOS DE REGISTRO

In classic bioelectric signal treatment different preprocessing techniques have been used for removing noise which is present in the electronic record, improving the signal to noise ratio (SNR). Microelectrode recordings (MER) are no stationary signals, compound by action potential sequences coming from the neural activity, background noise and artifacts. Different MER analysis methods reduce the background noise and the artifacts for the posterior classification. This article presents an awesome filtering procedure based in wavelets, that allows to establish that the background neural activity close to the electrode gives useful information about the process.

Impulse Response Estimation of Linear Time-Invariant Systems Using Convolved Gaussian Processes and Laguerre Functions

This paper presents a novel method to estimate the impulse response function of Linear Time-Invariant systems from input-output data by means of Laguerre functions and Convolved Gaussian Processes. We define a new non-stationary covariance function that encodes the convolution between the Laguerre functions and the input. The input (excitation) is modelled by a Gaussian Process prior. Thus, we are able to estimate the system’s impulse response by performing maximum likelihood estimation over the model hyperparameters. Besides, the proposed model performs well in missing and noisy data scenarios.

Indian Buffet Process for Model Selection in Latent Force Models

Latent force models (LFM) are an hybrid approach which combines multiple output Gaussian processes and differential equations, where the covariance functions encode the physical models given by the differential equations. LFM require the specification of the number of latent functions used to build the covariance function for the outputs. Furthermore, they assume that the output data is explained by using the entire set of latent functions, which is not the case in many real applications. We propose in this paper the use of an Indian Buffet process (IBP) as a way to perform model selection over the number of latent Gaussian processes in LFM applications. Furthermore, IBP allows us to infer the interconnection between latent functions and the outputs. We use variational inference to approximate the posterior distributions, and show examples of the proposed model performance over artificial data and a motion capture dataset.

COMPARACIÓN DE MÉTODOS DE CARACTERIZACIÓN DE SEÑALES MER

This document presents a microelectrode registers feature extraction methodologies comparison for brain zones identification found in Parkinson’s disease surgery. Best results are obtained using wavelet transforms, 97.37% and 71.4% for 2 and 4 classes, respectively.

SEGMENTACIÓN AUTOMÁTICA DE SEÑALES NO ESTACIONARIAS UTILIZANDO LA TRANSFORMADA WAVELET ESTACIONARIA

Bioelectrics signals carry events which deliver relevant information to specialists, depending on the analized signal. Due to physiological signals are characterized by events or abrupt changes (QRS complex in ECG, action potential in MER), the wavelet transform is employed to get a better representation of these events. Taking advantage of undecimated wavelet transform representation in the time-frequency plane we use abrupt change detection algorithms on the coefficients of each scale, for locating abrupt changes not just in the time plane but also in the frequency plane.

Control adaptivo por reubicación de polos

Se presentan los resultados de aplicar diferentes algoritmos de identificacion y control adaptivo a una planta de tercer orden a partir del modelo regresor de la misma. Entre los algoritmos de identificacion se incluyen proyeccion, proyeccion ortogonal, minimos cuadrados y filtro de Kalman. La tecnica de control utilizada es el control por reubicacion de polos. Se incluye el control un paso adelante y el control dead-beat como casos especificos del control por reubicacion de polos. Se muestra como el tiempo de muestreo se convierte en un parametro fundamental en el diseno del sistema. La mejor respuesta de la planta en lazo cerrado se obtiene usando el control dead-beat.

Segmentación automática de señales neurofisiológicas utilizando métodos basados en wavelets.

Bioelectric signals contain artifacts from a width variety of sources which brings within uncertainty to signal processing techniques and require stationary signal segments for its analysis. The present article develops a segmentation algorithm based on wavelets. The tests were done on parkinsonian patients database in thalamus and subthalamus brain regions drop as result that the proposed methodology lets to segment in an appropriate way.