Simple Independent Component Analysis Research Articles

A new method for QRS complex detection in multichannel ECG: Application to self-monitoring of fetal health

This paper proposes a new approach for QRS complex detection in multichannel ECG and presents its application to fetal QRS (fQRS) detection in signals acquired from maternal abdominal leads. The method exploits the characteristics of pseudo-periodicity and time shape of QRS, it consists of devising a quality index (QI) which synthesizes these characteristics and of finding the linear combination of the acquired ECGs, which maximizes this QI. In the application for fQRS detection two QIs are devised, one QI (mQI) for maternal ECG (mECG) and one QI (fQI) for fetal ECG (fECG). The method is completely unsupervised and based on the following steps: signal pre-processing; maternal QRS-enhanced signal extraction by finding the linear combination that maximize the mQI; detection of maternal QRSs; mECG component approximation and canceling by weighted Singular Value Decomposition (SVD); fQRS-enhanced signal extraction by finding the linear combination that maximize the fQI and fQRS detection. The proposed method was compared with our previously developed Independent Component Analysis (ICA) based method as well as with simple mECG canceling and simple ICA methods. The comparison was carried out by evaluating the performances of the procedures in fQRS detection. The new method outperformed the results of the other approaches on the annotated open set of the Computing in Cardiology Challenge 2013 database. The proposed method seems to be promising for its implementation on portable device and for use in self-monitoring of fetal health in pregnant women.

Efficient coding of spectrotemporal binaural sounds leads to emergence of the auditory space representation

To date a number of studies have shown that receptive field shapes of early sensory neurons can be reproduced by optimizing coding efficiency of natural stimulus ensembles. A still unresolved question is whether the efficient coding hypothesis explains formation of neurons which explicitly represent environmental features of different functional importance. This paper proposes that the spatial selectivity of higher auditory neurons emerges as a direct consequence of learning efficient codes for natural binaural sounds. Firstly, it is demonstrated that a linear efficient coding transform—Independent Component Analysis (ICA) trained on spectrograms of naturalistic simulated binaural sounds extracts spatial information present in the signal. A simple hierarchical ICA extension allowing for decoding of sound position is proposed. Furthermore, it is shown that units revealing spatial selectivity can be learned from a binaural recording of a natural auditory scene. In both cases a relatively small subpopulation of learned spectrogram features suffices to perform accurate sound localization. Representation of the auditory space is therefore learned in a purely unsupervised way by maximizing the coding efficiency and without any task-specific constraints. This results imply that efficient coding is a useful strategy for learning structures which allow for making behaviorally vital inferences about the environment.

A Simple and Accurate ICA Algorithm for Separating Mixtures of Up to Four Independent Components

This paper introduces an algorithm for independent component analysis (ICA) using explicit closed forms of two-, three-and four-dimensional antisymmetric matrix exponentials, based on which both the search direction and matrix exponentials can be directly computed in each iteration without any approximation. In addition, two errors have been corrected for the representation of four-dimensional antisymmetric matrix exponentials that were established in other works. Simulations show that the algorithm converges fast and can achieve better performance than the well-known Extended InfoMax and FastICA algorithms for mixtures of up to four independent components.

Reconsideration of the True versus Apparent Arctic Oscillation

Abstract The physical reality of the Arctic Oscillation (AO; or northern annular mode) is considered. The data used are mainly the monthly mean sea level pressure (SLP). A schematic figure is first presented to illustrate the relationship between the North Atlantic Oscillation (NAO)–Pacific–North American Oscillation (PNA) system and the AO–negative correlation mode between the Atlantic and the Pacific (AO–NCM) system. Although the NAO–PNA (apparent AO–NCM) and true AO–NCM systems give rise to the same EOFs, the probability density functions for the time coefficients of the two leading modes are different. Therefore, the discrimination of the two systems is possible. Several pieces of evidence indicate that, in the real world, the NAO–PNA and the AO–NCM are located on almost the same plane in phase space. This means that the NAO–PNA and AO–NCM systems have the same variations on the plane in common, implying that when the NAO–PNA system is real, the AO–NCM is unlikely to be real. Simple independent component analysis is carried out to distinguish between the true and apparent AO–NCM systems, indicating that the NAO and PNA are independent oscillations, that is, true ones. The analysis is extended to the winter mean SLP field, for which the EOF shows the NAO–PNA but not the AO–NCM. This may be due to the fact that the winter mean NAO and PNA patterns have little spatial correlation. Calculations using randomly selected samples also indicate that when the NAO and PNA patterns have little spatial correlation, the AO never appears as EOF1. All the preceding results show that almost all characteristics of the AO–NCM can be explained from those of the NAO–PNA. Hence it is concluded that the AO, which is extracted by EOF analysis from the temporarily independent but spatially overlapping variations of the NAO and PNA, is almost apparent.

Improving the Performance of ICA Algorithm for fMRI Simulated Data Analysis Using Temporal and Spatial Filters in the Preprocessing Phase

Introduction: The accuracy of analyzing Functional MRI (fMRI) data is usually decreases in the presence of noise and artifact sources. A common solution in for analyzing fMRI data having high noise is to use suitable preprocessing methods with the aim of data denoising. Some effects of preprocessing methods on the parametric methods such as general linear model (GLM) have previously been evaluated. In this study, besides the comparison of simple and noisy Independent Component Analysis (ICA) algorithms, the quantity effects of some spatial and temporal filtering have been evaluated on the functionality of ICA algorithms. Noisy ICA algorithms perform with a higher accuracy (up to 16%) in comparison to simple ICA for noisy fMRI data, although it is more time consuming than simple ICA. The accuracy of the results is improved by 8-10% using spatial and temporal filtering prior to simple ICA. Materials and Methods: Simple ICA and noisy ICA methods have been compared for analyzing simulated fMRI data sets. The impact of some temporal and spatial filters on the functionality of simple ICA algorithms has been evaluated. Implemented filters have been proposed in low and high pass group. Results: The sensitivity, specificity and temporal accuracy of simple ICA algorithms has been improved by using high pass filters. Although low pass filtering has some positive effects on the performance of simple ICA algorithms in the low SNR levels, in the high signal-noise Ratio (SNR) levels these low pass filters may cause a decrease in the sensitivity, specificity and temporal accuracy of simple ICA methods. Discussion and Conclusion: The results obtained from simple and noisy ICA algorithms for analyzing fMRI data having high SNR levels are approximately similar. Infomax algorithm uses Gradient based methods for estimating unmixing matrix has better sensitivity, specificity and temporal accuracy than Fast ICA for analyzing noisy ICA data. An alternative to the complicated and time consuming noisy ICA algorithms is to preprocess and denoise fMRI data prior to analyzing it by simple ICA algorithms.

Nested Newton's method for ICA and post factor analysis

Two distinct topics are dealt with. First, a new method for independent component analysis (ICA) has been constructed that exploits the invariance of criteria under component-wise scaling, which is intrinsic to ICA. This practical and simple ICA method is called the nested Newton's method. When the number of the channel of observation is less than a certain level, factor analysis (FA) is ineffective (bound for FA). The target of this paper is these cases. Three of many concrete advantages of the nested Newton's method are addressed. i) It is robust against Gaussian noise and outperforms existing methods, such as JADE and Fast ICA, especially under Gaussian noise conditions. ii) It is highly stable globally. iii) Each step resolves itself into two-dimensional (2-D) matrix problems. There is thus no need to deal with gigantic matrices, which means that fewer computational resources are required. Second, a method called "post factor analysis (post-FA)" is described that is aimed to be useful as post-processing for ICA. Although it is functionally similar to conventional FA, post-FA is a completely new method and is more powerful than conventional FA in compensation for its stronger assumption that there are mutually independent sources behind observations. By fully making use of this assumption, post-FA is capable of estimating the noise variance beyond the known limit for FA. Furthermore, it improves the accuracy of ICA to a considerable extent. Any ICA algorithm without prewhitening (pre-WH) or pre-factor-analysis (pre-FA) can be used for preprocessing, although the nested method is a good candidate.