A Local Learning Rule for Independent Component Analysis.
Ontology highlight
ABSTRACT:
SUBMITTER: Isomura T
PROVIDER: S-EPMC4914970 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:We developed a biologically plausible unsupervised learning algorithm, error-gated Hebbian rule (EGHR)-?, that performs principal component analysis (PCA) and independent component analysis (ICA) in a single-layer feedforward neural network. If parameter ??=?1, it can extract the subspace that major principal components span similarly to Oja's subspace rule for PCA. If ??=?0, it can separate independent sources similarly to Bell-Sejnowski's ICA rule but without requiring the same number of input and output neurons. Unlike these engineering rules, the EGHR-? can be easily implemented in a biological or neuromorphic circuit because it only uses local information available at each synapse. We analytically and numerically demonstrate the reliability of the EGHR-? in extracting and separating major sources given high-dimensional input. By adjusting ?, the EGHR-? can extract sources that are missed by the conventional engineering approach that first applies PCA and then ICA. Namely, the proposed rule can successfully extract hidden natural images even in the presence of dominant or non-Gaussian noise components. The results highlight the reliability and utility of the EGHR-? for large-scale parallel computation of PCA and ICA and its future implementation in a neuromorphic hardware.
| S-EPMC5789861 | biostudies-other
Project description:Although models based on independent component analysis (ICA) have been successful in explaining various properties of sensory coding in the cortex, it remains unclear how networks of spiking neurons using realistic plasticity rules can realize such computation. Here, we propose a biologically plausible mechanism for ICA-like learning with spiking neurons. Our model combines spike-timing dependent plasticity and synaptic scaling with an intrinsic plasticity rule that regulates neuronal excitability to maximize information transmission. We show that a stochastically spiking neuron learns one independent component for inputs encoded either as rates or using spike-spike correlations. Furthermore, different independent components can be recovered, when the activity of different neurons is decorrelated by adaptive lateral inhibition.
| S-EPMC2858697 | biostudies-literature
Project description:Independent component analysis (ICA) is a popular blind source separation technique used in many scientific disciplines. Current ICA approaches have focused on developing efficient algorithms under specific ICA models, such as instantaneous or convolutive mixing conditions, intrinsically assuming temporal independence or autocorrelation of the sources. In practice, the true model is not known and different ICA algorithms can produce very different results. Although it is critical to choose an ICA model, there has not been enough research done on evaluating mixing models and assumptions, and how the associated algorithms may perform under different scenarios. In this paper, we investigate the performance of multiple ICA algorithms under various mixing conditions. We also propose a convolutive ICA algorithm for echoic mixing cases. Our simulation studies show that the performance of ICA algorithms is highly dependent on mixing conditions and temporal independence of the sources. Most instantaneous ICA algorithms fail to separate autocorrelated sources, while convolutive ICA algorithms depend highly on the model specification and approximation accuracy of unmixing filters.
| S-EPMC4309392 | biostudies-literature
Project description:Independent Component Analysis (ICA) offers an effective data-driven approach for blind source extraction encountered in many signal and image processing problems. Although many ICA methods have been developed, they have received relatively little attention in the statistics literature, especially in terms of rigorous theoretical investigation for statistical inference. The current paper aims at narrowing this gap and investigates the statistical sampling properties of the colorICA (cICA) method. The cICA incorporates the correlation structure within sources through parametric time series models in the frequency domain and outperforms several existing ICA alternatives numerically. We establish the consistency and asymptotic normality of the cICA estimates, which then enables statistical inference based on the estimates. These asymptotic properties are further validated using simulation studies.
| S-EPMC7641017 | biostudies-literature
Project description:In the independent component model, the multivariate data are assumed to be a mixture of mutually independent latent components. The independent component analysis (ICA) then aims at estimating these latent components. In this article, we study an ICA method which combines the use of linear and quadratic autocorrelations to enable efficient estimation of various kinds of stationary time series. Statistical properties of the estimator are studied by finding its limiting distribution under general conditions, and the asymptotic variances are derived in the case of ARMA-GARCH model. We use the asymptotic results and a finite sample simulation study to compare different choices of a weight coefficient. As it is often of interest to identify all those components which exhibit stochastic volatility features we suggest a test statistic for this problem. We also show that a slightly modified version of the principal volatility component analysis can be seen as an ICA method. Finally, we apply the estimators in analysing a data set which consists of time series of exchange rates of seven currencies to US dollar. Supporting information including proofs of the theorems is available online.
| S-EPMC7266430 | biostudies-literature
Project description:Independent Component Analysis is a matrix factorization method for data dimension reduction. ICA has been widely applied for the analysis of transcriptomic data for blind separation of biological, environmental, and technical factors affecting gene expression. The study aimed to analyze the publicly available esophageal cancer data using the ICA for identification and comprehensive analysis of reproducible signaling pathways and molecular signatures involved in this cancer type. In this study, four independent esophageal cancer transcriptomic datasets from GEO databases were used. A bioinformatics tool « BiODICA-Independent Component Analysis of Big Omics Data» was applied to compute independent components (ICs). Gene Set Enrichment Analysis (GSEA) and ToppGene uncovered the most significantly enriched pathways. Construction and visualization of gene networks and graphs were performed using the Cytoscape, and HPRD database. The correlation graph between decompositions into 30 ICs was built with absolute correlation values exceeding 0.3. Clusters of components-pseudocliques were observed in the structure of the correlation graph. The top 1,000 most contributing genes of each ICs in the pseudocliques were mapped to the PPI network to construct associated signaling pathways. Some cliques were composed of densely interconnected nodes and included components common to most cancer types (such as cell cycle and extracellular matrix signals), while others were specific to EC. The results of this investigation may reveal potential biomarkers of esophageal carcinogenesis, functional subsystems dysregulated in the tumor cells, and be helpful in predicting the early development of a tumor.
| S-EPMC8594933 | biostudies-literature
Project description:The independent component analysis (ICA) based methods are among the most prevalent techniques used for non-invasive fetal electrocardiogram (NI-fECG) processing. Often, these methods are combined with other methods, such adaptive algorithms. However, there are many variants of the ICA methods and it is not clear which one is the most suitable for this task. The goal of this study is to test and objectively evaluate 11 variants of ICA methods combined with an adaptive fast transversal filter (FTF) for the purpose of extracting the NI-fECG. The methods were tested on two datasets, Labour dataset and Pregnancy dataset, which contained real records obtained during clinical practice. The efficiency of the methods was evaluated from the perspective of determining the accuracy of detection of QRS complexes through the parameters of accuracy (ACC), sensitivity (SE), positive predictive value (PPV), and harmonic mean between SE and PPV (F1). The best results were achieved with a combination of FastICA and FTF, which yielded mean values of ACC = 83.72%, SE = 92.13%, PPV = 90.16%, and F1 = 91.14%. Time of calculation was also taken into consideration in the methods. Although FastICA was ranked to be the sixth fastest with its mean computation time of 0.452 s, it had the best ratio of performance and speed. The combination of FastICA and adaptive FTF filter turned out to be very promising. In addition, such device would require signals acquired from the abdominal area only; no need to acquire reference signal from the mother’s chest.
| S-EPMC10243647 | biostudies-literature
Project description:The present study was aimed to evaluate resting-state functional connectivity and topological properties of brain networks in narcolepsy patients compared with healthy controls. Resting-state fMRI was performed in 26 adult narcolepsy patients and 30 matched healthy controls. MRI data were first analyzed by group independent component analysis, then a graph theoretical method was applied to evaluate the topological properties in the whole brain. Small-world network parameters and nodal topological properties were measured. Altered topological properties in brain areas between groups were selected as region-of-interest seeds, then the functional connectivity among these seeds was compared between groups. Partial correlation analysis was performed to evaluate the relationship between the severity of sleepiness and functional connectivity or topological properties in the narcolepsy patients. Twenty-one independent components out of 48 were obtained. Compared with healthy controls, the narcolepsy patients exhibited significantly decreased functional connectivity within the executive and salience networks, along with increased functional connectivity in the bilateral frontal lobes within the executive network. There were no differences in small-world network properties between patients and controls. The altered brain areas in nodal topological properties between groups were mainly in the inferior frontal cortex, basal ganglia, anterior cingulate, sensory cortex, supplementary motor cortex, and visual cortex. In the partial correlation analysis, nodal topological properties in the putamen, anterior cingulate, and sensory cortex as well as functional connectivity between these regions were correlated with the severity of sleepiness (sleep latency, REM sleep latency, and Epworth sleepiness score) among narcolepsy patients. Altered connectivity within the executive and salience networks was found in narcolepsy patients. Functional connection changes between the left frontal cortex and left caudate nucleus may be one of the parameters describing the severity of narcolepsy. Changes in the nodal topological properties in the left putamen and left posterior cingulate, changes in functional connectivity between the left supplementary motor area and right occipital as well as in functional connectivity between the left anterior cingulate gyrus and bilateral postcentral gyrus can be considered as a specific indicator for evaluating the severity of narcolepsy.
| S-EPMC6616568 | biostudies-literature
Project description:The objective of this study was to test, in single subjects, the hypothesis that the signs of voluntary movement-related neural activity would first appear in the prefrontal region, then move to both the medial frontal and posterior parietal regions, progress to the medial primary motor area, lateralize to the contralateral primary motor area and finally involve the cerebellum (where feedback-initiated error signals are computed). Six subjects performed voluntary finger movements while DC coupled EEG was recorded from 64 scalp electrodes. Event-related potentials (ERPs) averaged on the movements were analysed both before and after independent component analysis (ICA) combined with dipole source analysis (DSA) of the independent components. Both a simple topographic analysis of undecomposed ERPs and the ICA/DSA analysis suggested that the original hypothesis was inadequate. The major departure from its predictions was that, while activity over many brain regions did appear at the expected times, it also appeared at unexpected times. Overall, the results suggest that the neuroscientific 'standard model', in which neural activity occurs sequentially in a series of discrete local areas each specialized for a particular function, may reflect the true situation less well than models in which large areas of brain shift simultaneously into and out of common activity states.
| S-EPMC2289049 | biostudies-literature
Project description:A method of analysis is presented that allows for the separation of specific radiation-induced changes into distinct components in real space. The method relies on independent component analysis (ICA) and can be effectively applied to electron density maps and other types of maps, provided that they can be represented as sets of numbers on a grid. Here, for glucose isomerase crystals, ICA was used in a proof-of-concept analysis to separate temperature-dependent and temperature-independent components of specific radiation-induced changes for data sets acquired from multiple crystals across multiple temperatures. ICA identified two components, with the temperature-independent component being responsible for the majority of specific radiation-induced changes at temperatures below 130 K. The patterns of specific temperature-independent radiation-induced changes suggest a contribution from the tunnelling of electron holes as a possible explanation. In the second case, where a group of 22 data sets was collected on a single thaumatin crystal, ICA was used in another type of analysis to separate specific radiation-induced effects happening on different exposure-level scales. Here, ICA identified two components of specific radiation-induced changes that likely result from radiation-induced chemical reactions progressing with different rates at different locations in the structure. In addition, ICA unexpectedly identified the radiation-damage state corresponding to reduced disulfide bridges rather than the zero-dose extrapolated state as the highest contrast structure. The application of ICA to the analysis of specific radiation-induced changes in real space and the data pre-processing for ICA that relies on singular value decomposition, which was used previously in data space to validate a two-component physical model of X-ray radiation-induced changes, are discussed in detail. This work lays a foundation for a better understanding of protein-specific radiation chemistries and provides a framework for analysing effects of specific radiation damage in crystallographic and cryo-EM experiments.
| S-EPMC5829680 | biostudies-literature