A bstract The unpredictability of seizures is one the most compromising features reported by people with epilepsy. Non-stigmatizing and easy-to-use wearable devices may provide information to predict based on physiological data. We propose a patient-agnostic seizure prediction method that identifies group-level patterns across data from multiple patients. employ supervised long-short-term networks (LSTMs) add unsupervised deep canonically correlated autoencoders (DCCAE) 24-hour using...

Physical exercise has been shown to modulate activity within the autonomic nervous system (ANS). Considering physical as a holistic stimulus on and specifically ANS, uni- multimodal analysis tools were applied characterize centrally driven interactions control of ANS functions. 19 young physically active participants performed treadmill tests at individually determined moderate high intensities. Continuous electrodermal (EDA), heart rate (HR), skin temperature wrist (Temp) recorded by...

Deep correlation-based multiview representation learning techniques have become increasingly popular methods for extracting highly correlated representations from data. However, their ability to find complex mappings between the views can also lead overfitting and overly representations. In this work, we propose a regularizer specific problem, based on Rademacher complexity of DNNs, tailored correlation maximization. We demonstrate that proposed regularization leads less noisy in synthetic...

Data fusion—the joint analysis of multiple datasets—through coupled factorizations has the promise to enable enhanced knowledge discovery, and hence is an active area. Various formulations matrix have been proposed, each with its own modeling assumptions. In this paper, we study two such methods, namely Independent Vector Analysis (IVA), i.e., extension Component (ICA) datasets, PARAFAC2, a tensor factorization approach. We demonstrate assumptions IVA PARAFAC2 using simulations, revealing...

This paper addresses the problem of detecting number signals correlated across multiple data sets with small sample support. While there have been studies involving two sets, more than has less explored. In this work, a rank-reduced hypothesis test for is presented scenarios where samples compared to dimensions sets.

This paper presents a detection scheme for determining the number of signals that are correlated across multiple data sets when sample size is small compared to dimensions sets. To accommodate sample-poor regime, we decouple problem into several independent two-channel order-estimation problems may be solved separately by combination principal component analysis (PCA) and canonical correlation (CCA). Since all must subset between any pair sets, keep only each Then, criterion inspired...

Purpose: Running an ultramarathon can be considered as a multifaceted, intense stressor inducing changes within the autonomic nervous system (ANS). The aim of this study was to examine and across ANS modalities in response ultramarathon.Methods: Thirteen runners (44.3 ± 5.9 years) completed 65 km run. Electrodermal activity (EDA), heart rate (HR), skin temperature measured at wrist (Temp), were recorded before after running. Three-minute intervals analysed. Mean values compared by t-tests...

We present a scheme for determining the number of signals common to or correlated across multiple data sets. Handling sets is challenging due different possible correlation structures. For two sets, are either uncorrelated between however, there numerous combinations how can be correlated. Prior studies dealing with all assume particular structure. In this paper, we technique based on series hypothesis tests and bootstrap, which works arbitrary Numerical results show that proposed correctly...

Deep canonical correlation analysis (DCCA) is often applied to paired data samples from diverse sources extract meaningful common information. However, when the are heterogeneous, some of useful information may be complementary but not exactly common. In spite this fact, existing techniques learn maximally correlated representations multiple views and formulated so that they aim yield identical latent subspaces for each view. This approach sub-optimal in estimating true signal heterogeneous...

Traditional model-order selection for canonical correlation analysis infers latent correlations between two sets of noisy data. In this scenario it is enough to count the number correlated signals, and thus model order a scalar. When problem generalized collection three or more data sets, signals can demonstrate all some subset, one cannot completely describe structure. We present method estimating multiset structure that combines source extraction in style joint blind separation with...

Identifying relationships among multiple datasets is an effective way to summarize information and has been growing in importance. In this paper, we propose a robust 3-step method for identifying the relationship structure based on Independent Vector Analysis (IVA) bootstrap-based hypothesis testing. Unlike previous approaches, our theory-backed eliminates need user-defined thresholds can effectively handle non-Gaussian data. It achieves by incorporating higher-order statistics through IVA...

A complex-valued signal is improper if it correlated with its complex conjugate. The dimension of the subspace, i.e., number components in a measurement, an important parameter and unknown most applications. In this letter, we introduce two approaches to estimate dimension: one based on information-theoretic criterion other hypothesis testing. We also present reduced-rank versions these that work for scenarios where observations comparable or even smaller than data. Unlike techniques...

This paper is concerned with the analysis of correlation between two high-dimensional data sets when there are only few correlated signal components but number samples very small, possibly much smaller than dimensions data. In such a scenario, principal component (PCA) rank-reduction preprocessing step commonly performed before applying canonical (CCA). We present simple, yet effective approaches to joint model-order selection that should be retained through PCA and signals. These based on...

We propose a robust technique for multi-speaker voice activity detection and source enumeration in wireless acoustic sensor networks (WASN). The proposed first clusters the nodes that observe single speaker as dominant source, then estimates of each by introducing block-sparsity penalizing term unmixing problem. method is scalable terms number simultaneously active speakers, does not require setting empirical thresholds, to impulsive noise sources. results are validated using WASN with four...

Estimating the number of correlated components between two data sets is a challenging task in case small sample support. Typically, rank-reduction preprocessing step based on principal component analysis (PCA) carried out each set individually to reduce dimensionality before analyzing correlation sets. However, PCA retains with largest variance within set, and therefore fails when these are not ones that account for To overcome this, we propose an alternative technique that, instead...

The identification of the dependent components in multiple data sets is a fundamental problem many practical applications. challenge these applications that often are high-dimensional with few observations or available samples and contain latent unknown probability distributions. A novel mathematical formulation this proposed, which enables inference underlying correlation structure strict false positive control. In particular, discovery rate controlled at pre-defined threshold on two levels...

Detecting the components common or correlated across multiple data sets is challenging due to a large number of possible correlation structures among components. Even more determine precise structure these correlations. Traditional work has focused on determining only model order, i.e., dimension subspace, that depends how model-order problem defined. Moreover, identifying order often not enough understand relationship in different sets. We aim at solving complete modelselection problem,...