Allan Kardec Barros

MSE behaviour of biomedical event-related filters [impedance cardiography application]

The mean-squared error (MSE) behaviour for Fourier linear combiner (FLC)-based filters is analyzed, using the independence assumption. The advantage of this analysis is its simplicity compared with previous results. The MSE transient behaviour for this kind of filters is also presented for the first time. Moreover, a time-varying sequence for the least mean square (LMS) algorithm step-size is proposed to provide fast convergence with small misadjustment error. It is shown that for this sequence, the MSE behaves better as the input signal-to-noise ratio (SNR) decreases, but increases with the number of harmonics. Lastly, the authors make a brief analysis on the nonstationary behaviour of these filters, and again they find simple expressions for the MSE behaviour.

Extraction of Sleep-Spindles from the Electroencephalogram (EEG)

Independent component analysis (ICA) is a powerful tool for separating signals from their observed mixtures. This area of research has produced many varied algorithms and approaches to the solution of this problem. The majority of these methods adopt a truly blind approach and disregard available a priori information in order to extract the original sources or a specific desired signal. In this contribution we propose a fixed point algorithm which utilizes a priori information in finding a specified signal of interest from the sensor measurements. This technique is applied to the extraction and channel isolation of sleep spindles from a multi-channel electroencephalograph (EEG).

An algorithm based on the even moments of the error

We propose an algorithm based on a linear combination of the even moments of the error for adaptive filtering, called weighted even moment (WEM) algorithm. It is similar to the well-known least mean square (LMS) and to the family of algorithms proposed by Walach and Widrow (1994). This later ones were shown to behave poorer than the LMS, however, when the noise was Gaussian. We study the WEM algorithm convergence behavior and deduce equations for the misadjustment and the learning time. The results showed that the WEM had better performance than the LMS when the noise had a Gaussian distribution.

Pre-filtering non-stationary signals to improve blind source separation

Much work has been carried out concerning blind source separation (BSS) that uses the principle of independent component analysis (ICA), which stands as an elegant, simple and powerful tool to BSS. However, the algorithms previously proposed are still sensitive to nonstationarity, which alters in a bad sense the separation. We propose here a simple pre-processing to the mixed signals based on low-pass filtering that ensures a much better performance for the ICA algorithms.

Analysis of the Time Constant for the Sigmoidal Algorithm Applied to Biomedical Signals

The time constant along with misadjustment offer a manner of analyzing the convergence behavior of adaptive algorithms, which are largely used to process biomedical signals, mainly regular ones such as cardiac signals. However, some equations found for the time constant suggested in the literature are noise dependent, yielding infinite value for the noiseless case, which is obviously wrong. This problem may explain the fact that no actual comparison of the found equations to practically derived time constant were shown in the literature. In this work we analyze the time constant for updating the weights for the sigmoidal algorithm (SA) when it is used to carry out adaptive estimation of the cardiac component of the impedance cardiographic (ICG) signal

Real-time separation of mixed non-stationary signals

In this work, the authors describe an algorithm for the real-time separation of mixed non-stationary signals. The algorithm is based on entropy maximization of the system output, and the final recursive updating equation is simple and easy to implement.

Comparison of saccade-related EEG signal with saccade-related independent component

We have been research saccade-related EEG signals in order to predict beginning of saccade by EEG signal. We have already detected saccade-related EEG signal by ensemble averaging and saccade-related independent components (ICs) by independent component analysis (ICA). However, features of saccade-related EEG signals and saccade-related ICs were not compared. In this paper, saccade-related EEG signals and saccade-related ICs were compared in the point of the latency between starting time of a saccade and time when a saccade-related EEG signal or an IC has maximum value and in the point of the peak scale where a saccade-related EEG signal or an IC has maximum value

Neural Coding by Temporal and Spatial Correlations

Redundancy reduction as a form of neural coding has been since the early sixties a topic of large research interest. A number of strategies has been proposed, but the one which is attracting most attention recently assumes that this coding is carried out so that the output signals are mutually independent. In this work we go one step further and suggest an algorithm that separates also non-orthogonal signals (i.e., “dependent” signals). The resulting algorithm is very simple, as it is computationally economic and based on second order statistics that, as it is well know, is more robust to errors than higher order statistics, moreover, the permutation/scaling problem (Comon, 1994) is avoided. The framework is given with a biological background, as we avocate throughout the manuscript that the algorithm fits well the single neuron and redundancy reduction doctrine, but it can be used as well in other applications such as biomedical engineering and telecommunications.