Search results for "Matrix decomposition"
showing 10 items of 43 documents
LogDet divergence-based metric learning with triplet constraints and its applications.
2014
How to select and weigh features has always been a difficult problem in many image processing and pattern recognition applications. A data-dependent distance measure can address this problem to a certain extent, and therefore an accurate and efficient metric learning becomes necessary. In this paper, we propose a LogDet divergence-based metric learning with triplet constraints (LDMLT) approach, which can learn Mahalanobis distance metric accurately and efficiently. First of all, we demonstrate the good properties of triplet constraints and apply it in LogDet divergence-based metric learning model. Then, to deal with high-dimensional data, we apply a compressed representation method to learn…
Measuring High-Order Interactions in Rhythmic Processes Through Multivariate Spectral Information Decomposition
2021
Many complex systems in physics, biology and engineering are modeled as dynamical networks and described using multivariate time series analysis. Recent developments have shown that the emergent dynamics of a network system are significantly affected by interactions involving multiple network nodes which cannot be described using pairwise links. While these higher-order interactions can be probed using information-theoretic measures, a rigorous framework to describe them in the frequency domain is still lacking. This work presents an approach for the spectral decomposition of multivariate information measures, capable of identifying higher-order synergistic and redundant interactions betwee…
Diagnosis of Sensor Faults in PMSM and Drive System Based on Structural Analysis
2021
This paper presents a model-based fault diagnosis method to detect sensor faults in permanent magnet synchronous motor (PMSM) drives based on structural analysis technique. The structural model is built based on the dynamic model of the PMSM in matrix form, including unknown variables, known variables, and faults. The Dulmage-Mendelsohn (DM) decomposition is applied to evaluate the redundancy of the model and obtain redundant testable sub-models. These testable redundant sub-models are used to form residuals to observe the system state, and distinguish between healthy and faulty conditions. This work investigates faults in eleven sensors in a PMSM drive, thus nine structured residuals are d…
Quantitative evaluation of muscle synergy models: a single-trial task decoding approach.
2012
Delis, Ioannis | Berret, Bastien | Pozzo, Thierry | Panzeri, Stefano; International audience; ''Muscle synergies, i.e., invariant coordinated activations of groups of muscles, have been proposed as building blocks that the central nervous system (CNS) uses to construct the patterns of muscle activity utilized for executing movements . Several efficient dimensionality reduction algorithms that extract putative synergies from electromyographic (EMG) signals have been developed. Typically, the quality of synergy decompositions is assessed by computing the Variance Accounted For (VAF). Yet, little is known about the extent to which the combination of those synergies en codes task discriminating…
Listwise Recommendation Approach with Non-negative Matrix Factorization
2018
Matrix factorization (MF) is one of the most effective categories of recommendation algorithms, which makes predictions based on the user-item rating matrix. Nowadays many studies reveal that the ultimate goal of recommendations is to predict correct rankings of these unrated items. However, most of the pioneering efforts on ranking-oriented MF predict users’ item ranking based on the original rating matrix, which fails to explicitly present users’ preference ranking on items and thus might result in some accuracy loss. In this paper, we formulate a novel listwise user-ranking probability prediction problem for recommendations, that aims to utilize a user-ranking probability matrix to predi…
Spectral WENO schemes with Adaptive Mesh Refinement for models of polydisperse sedimentation
2012
The sedimentation of a polydisperse suspension with particles belonging to N size classes (species) can be described by a system of N nonlinear, strongly coupled scalar first-order conservation laws. Its solutions usually exhibit kinematic shocks separating areas of different composition. Based on the so-called secular equation [J. Anderson, Lin. Alg. Appl. 246, 49–70 (1996)], which provides access to the spectral decomposition of the Jacobian of the flux vector for this class of models, Burger et al. [J. Comput. Phys. 230, 2322–2344 (2011)] proposed a spectral weighted essentially non-oscillatory (WENO) scheme for the numerical solution of the model. It is demonstrated that the efficiency …
Quantifying brain tumor tissue abundance in HR-MAS spectra using non-negative blind source separation techniques
2012
Given high-resolution magic angle spinning (HR-MAS) spectra from several glial tumor subjects, our goal is to differentiate between tumor tissue types by separating the different sources that contribute to the profile of each spectrum. Blind source separation techniques are applied for obtaining characteristic profiles for necrosis, highly cellular tumor and border tumor tissue and providing the contribution (abundance) of each of these tumor tissue types to the profile of each spectrum. The problem is formulated as a non-negative source separation problem. Non-negative matrix factorization, convex analysis of non-negative sources and non-negative independent component analysis methods are …
Ray-Space-Based Multichannel Nonnegative Matrix Factorization for Audio Source Separation
2021
Nonnegative matrix factorization (NMF) has been traditionally considered a promising approach for audio source separation. While standard NMF is only suited for single-channel mixtures, extensions to consider multi-channel data have been also proposed. Among the most popular alternatives, multichannel NMF (MNMF) and further derivations based on constrained spatial covariance models have been successfully employed to separate multi-microphone convolutive mixtures. This letter proposes a MNMF extension by considering a mixture model with Ray-Space-transformed signals, where magnitude data successfully encodes source locations as frequency-independent linear patterns. We show that the MNMF alg…
Tensor decomposition of EEG signals: A brief review
2015
Electroencephalography (EEG) is one fundamental tool for functional brain imaging. EEG signals tend to be represented by a vector or a matrix to facilitate data processing and analysis with generally understood methodologies like time-series analysis, spectral analysis and matrix decomposition. Indeed, EEG signals are often naturally born with more than two modes of time and space, and they can be denoted by a multi-way array called as tensor. This review summarizes the current progress of tensor decomposition of EEG signals with three aspects. The first is about the existing modes and tensors of EEG signals. Second, two fundamental tensor decomposition models, canonical polyadic decomposit…
A new Frequency Domain Measure of Causality based on Partial Spectral Decomposition of Autoregressive Processes and its Application to Cardiovascular…
2019
We present a new method to quantify in the frequency domain the strength of directed interactions between linear stochastic processes. This issue is traditionally addressed by the directed coherence (DC), a popular causality measure derived from the spectral representation of vector autoregressive (AR) processes. Here, to overcome intrinsic limitations of the DC when it needs to be objectively quantified within specific frequency bands, we propose an approach based on spectral decomposition, which allows to isolate oscillatory components related to the pole representation of the vector AR process in the Z-domain. Relating the causal and non-causal power content of these components we obtain…