Latent Variable Analysis and Signal Separation

Name: Latent Variable Analysis and Signal Separation | 10th International Conference, LVA/ICA 2012, Tel Aviv, Israel, March 12-15, 2012, Proceedings
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

10th International Conference, LVA/ICA 2012, Tel Aviv, Israel, March 12-15, 2012, Proceedings

Fabian Theis Andrzej Cichocki Arie Yeredor Michael Zibulevsky(Editor)

Springer (Publisher)

Published on 9. February 2012

XVI, 538 pages

E-Book

PDF with digital watermarking

System requirements

978-3-642-28551-6 (ISBN)

€53.49incl. 7% vat

System requirements

for PDF with digital watermarking

E-Book Single Licence

Available for download

Description

More details

Other editions

Content

Title page
Preface
Organization
Table of Contents
General LVA/ICA Theory, Methods and Extensions
Block Component Analysis, a New Concept for Blind Source Separation
Algebraic Tools
Block Component Analysis: The Concept
Illustration
Toy Example: Audio
Application in Wireless Communication
Discussion and Conclusion
References
Partially Linear Estimation with Application to Image Deblurring Using Blurred/Noisy Image Pairs
Introduction
Partially Linear Estimation
Application to Sparse Approximations
Image Deblurring with Blurred/Noisy Image Pairs
Conclusion
References
Causal Discovery for Linear Non-Gaussian Acyclic Models in the Presence of Latent Gaussian Confounders
Introduction
Related Works Concerning Latent Confounders
Causal Discovery for Causal Pairs with Latent Confounders
Cumulant-Based Measure by Aapo Hyvärinen
Cause-Effect Pairs in Presence of Latent Gaussian Confounders
Normalization to Unit Variance / Unit Absolute Kurtosis
New Cumulant-Based Measure
Experiment
Synthesis Data
Real World Data
Conclusion
References
Alleviating the Influence of Weak Data Asymmetries on Granger-Causal Analyses
Introduction
Methods
SISEC Challenge Simulated EEG Dataset
Granger Causality
Exploiting Statistical Characterics of Non-/interacting Signals for Assessing the Reliability Causal Predictions
A Test for Assessing the Time-Lagged Nature of Interactions
Experiments
Results
Discussion
Conclusion
References
Online PLCA for Real-Time Semi-supervised Source Separation
Introduction
Proposed Algorithm
Online Separation and Dictionary Learning
Mixture Frame Classification
Algorithm Summary
Experiments
Conclusions
References
Cram´er-Rao Bound for Circular Complex Independent Component Analysis
Introduction
Prerequisites
Complex Functions and Complex Random Vectors
Cramér-Rao Bound for a Complex Parameter
Derivation of Cramér-Rao Bound
CRB for G=WA
CRB for W
Results for Generalized Gaussian Distribution (GGD)
Conclusion
References
Complex Non-Orthogonal Joint Diagonalization Based on LU and LQ Decompositions
Introduction
Proposed Algorithms
Framework for the Proposed Algorithms
Schemes to Find Optimal Elementary Rotation Matrices
Remarks and Summarization
Simulation Results
Conclusion
References
Exact and Approximate Quantum Independent Component Analysis for Qubit Uncoupling
Introduction
Mixing Model
Separating System
Exact QICA
Approximate QICA
Extensions and Conclusions
References
A Matrix Joint Diagonalization Approach for Complex Independent Vector Analysis
Introduction
Problem Descriptions and Prelimiaries
Complex Independent Vector Analysis
Complex Oblique Projective Manifold
A CG Algorithm for Simultaneous Non-unitary SVD
Numerical Experiments
References
Algebraic Solutions to Complex Blind Source Separation
Introduction
Complex BSS and Second-Order Statistics
Complex Linear BSS Model
Second-Order Statistics Based Algebraic Solutions
Algebraic Solutions to Complex BSS Problem
Two Hermitian or Two Complex Symmetric
One Hermitian and One Complex Symmetric
Numerical Experiments
References
On the Separation Performance of the Strong Uncorrelating Transformation When Applied to Generalized Covariance and Pseudo-covariance Matrices
Introduction and Model Assumptions
Normalization Model and the SUT
Performance Analysis
Simulation
Conclusion
References
A Canonical Correlation Analysis Based Method for Improving BSS of Two Related Data Sets
Introduction
Our Method
Experimental Results
References
A Probability-Based Combination Method for Unsupervised Clustering with Application to Blind Source Separation
Introduction
The Proposed Combination Algorithm
The Basic Idea
The Clustering Algorithm
Application to Blind Source Separation
Hard-Decision Approach
Extension to Soft-Decision
Experimental Results
Conclusion
References
Charrelation Matrix Based ICA
Introduction
Charrelation Matrices
Derivation of the Separation Scheme
Results
Discussion
Appendix
References
Contrast Functions for Independent Subspace Analysis
Introduction
The ISA Problem
Deflationary Contrast Functions, ICA, and ISA
Contrast Functions
Contrasts and ISA
Sub- and Super-Gaussian Subspaces
Other Types of Norm Dependence
References
Distributional Convergence of Subspace Estimates in FastICA: A Bootstrap Study
Introduction
Materials and Methods
Results
Discussion
References
New Online EM Algorithms for General Hidden Markov Models. Application to the SLAM Problem
Introduction
New Online EM Algorithms for General HMM
Experiments
Conclusion
References
The Role of Whitening for Separation of Synchronous Sources
Introduction
Background
Phase-Locking Factor
Whitening
Condition Number
Upper Bound for Condition Number after Prewhitening
Notation and Assumptions
Upper Bound
Experiments
Discussion
Conclusion
References
Simultaneous Diagonalization of Skew-Symmetric Matrices in the Symplectic Group
Introduction
Interacting Sources, Skew-Symmetric Matrices and the Induced Symplectic Geometry
Optimizing the Symplectic Transformation
Simulations and Conclusion
References
Joint Block Diagonalization Algorithms for Optimal Separation of Multidimensional Components
Introduction
Derivation of the Relative Variations
Algorithms
Simulations
References
On Computation of Approximate Joint Block-Diagonalization Using Ordinary AJD
Introduction
Survey of Main AJD Methods
AJD Methods in the Block Scenario
Clustering of AJD Components
U-WEDGE Provides Perfect Separation of the Blocks
Simulation Experiments
Conclusions
References
Joint Diagonalization of Several Scatter Matrices for ICA
Introduction
Scatter Functionals
Independent Component (IC) Functionals
Joint Diagonalization of Several Scatter Functionals
Practical Implementation
Simulation
Conclusions
References
To Infinity and Beyond: On ICA over Hilbert Spaces
Introduction
Vector Spaces and Hilbert Spaces
Hilbert Spaces
Statistics on Hilbert Spaces
Random Variables on Hilbert Spaces
Independence of Infinitely Many Components
Moments
Characteristic Function
Separability of ICA on Hilbert Spaces
Discussion
References
Sparsity, Sparse Coding and Dictionary Learning
Regularized Sparse Representation for Spectrometric Pulse Separation and Counting Rate Estimation
Introduction
Methodology
Model Description
Overview on the group LASSO
Proposed Algorithm for Pileup Separation and Activity Estimation
Results and Discussion
Simulations Protocol
Results and Discussion
Conclusion
References
Some Uniqueness Results in Sparse Convolutive Source Separation
Introduction
Permutation and Scaling Ambiguities in Frequency Domain Filter Estimation
Exploiting Sparsity to Solve the Permutation Ambiguity
Main Result and Structure of the Paper
Theoretical Guarantees
Quantification of Permutations
Exploitation of an Uncertainty Principle
Combinatorial Arguments
Extending Theorem 2 to Non-prime Filter Length L?
Numerical Experiments
Proposed Combinatorial Algorithm
Monte-Carlo Simulations
Conclusions
References
Ternary Sparse Coding
Introduction
Sparse Coding with Ternary Hidden Variables
Numerical Experiments
Discussion
References
Closed-Form EM for Sparse Coding and Its Application to Source Separation
Introduction
Closed-Form EM for a Spike-and-Slab Sparse Coding Model
Numerical Experiments
Discussion
References
Convolutive Underdetermined Source Separation through Weighted Interleaved ICA and Spatio-temporal Source Correlation
Introduction
Estimation of the Mixing Parameters
Weighted Natural Gradient
Interleaved Orthogonal Adaptations
Permutation Alignment
TDOA Vector Estimation
Multi-resolution Spatio-temporal Correlation
Source Recovery
Experimental Results
Conclusion
References
Dictionary Learning with Large Step Gradient Descent for Sparse Representations
Introduction
Dictionary Learning
Problem
Algorithms
Motivations for an Adaptive Gradient Step Size
Identifying the Global Optimum: Learning with a Fixed Support
Empirical Observations on Existing Algorithms
Large Step Gradient Descent
Optimal Step Projected Gradient Descent
Experimental Validation
Learning with a Fixed Support
Complete Learning
Conclusion
References
Separation of Sparse Signals in Overdetermined Linear-Quadratic Mixtures
Introduction
Mixing Model
Sparsity-Based Cancellation of Quadratic Terms
The main idea
Theoretical Aspects
Implementation Issues
Results
Conclusions
References
Collaborative Filtering via Group-Structured Dictionary Learning
Introduction
The OSDL Problem
OSDL Based Collaborative Filtering
Numerical Results
Evaluation
Conclusions
References
Group Polytope Faces Pursuit for Recovery of Block-Sparse Signals
Introduction
Review of the Polytope Faces Pursuit Algorithm
Recovery of Block-Sparse Signals via Group Polytope Faces Pursuit
Group Selection Criterion
Dual Linear Program of the Group Sparse Recovery Problem
The Proposed Algorithm
Simulation Results
Conclusions
References
Non-negative and Other Factorizations
Nonnegative Matrix Factorization via Generalized Product Rule and Its Application for Classification
Introduction
NMF via Generalized Product Rules
Implementation of u-NMF
u-NMF for Classification
Illustrative Example
Benchmark Tests
Conclusion
References
An Algebraic Method for Approximate Rank One Factorization of Rank Deficient Matrices
Introduction
Finding Joint Rank One Factors
Experiments
Conclusion
Appendix
References
Bayesian Non-negative Matrix Factorization with Learned Temporal Smoothness Priors
Introduction
Unsupervised Algorithms
NMF Framework
NMF-MU Algorithm
NMF-EM Algorithm
Bayesian NMF with Temporal Smoothness Prior
Supervised Algorithms
Evaluation
Conclusions and Perspectives
References
On Connection between the Convolutive and Ordinary Nonnegative Matrix Factorizations
Introduction
A Novel Derivation for CNMF Algorithms
Alternative Least Squares Algorithm for CNMF
Initialization for CNMF Algorithms
Simulations
Conclusions
References
On Revealing Replicating Structures in Multiway Data: A Novel Tensor Decomposition Approach
Problem Formulation
Notation and Basic Multilinear Algebra
Decomposition Methods
Orthogonal Patterns
Nonnegative Patterns
Simulations
Synthetic Data
Analysis of Texture Images
Analysis of Patterns inMusic
Conclusions
References
Audio Separation and Analysis
An On-Line NMF Model for Temporal Pattern Learning: Theory with Application to Automatic Speech Recognition
Introduction
Model
Time-Coded NMF
Learning
Decoding
Application to Speech Recognition
Discussion and Conclusion
References
Low-Latency Instrument Separation in Polyphonic Audio Using Timbre Models
Introduction
Spectral Bin Classification Masks
Harmonic Mask
Pitch Likelihood Estimation
Timbre Classification
Instrument Pitch Tracking
Evaluation
Conclusions
References
Real-Time Speech Separation by Semi-supervised Nonnegative Matrix Factorization
Introduction
Nonnegative Matrix Factorization (NMF) for Source Separation
On-Line NMF
On-Line Supervised NMF
On-Line Semi-supervised NMF
Real-Time Implementation
Experimental Evaluation
Experimental Settings
Results
Conclusion
References
An Audio-Video Based IVA Algorithm for Source Separation and Evaluation on the AV16.3 Corpus
Introduction
Independent Vector Analysis Based Methods
Model
Independent Vector Analysis
Fast Fixed-Point Independent Vector Analysis
Audio-Video Based Independent Vector Analysis
Pitch Difference Based Evaluation for Real Recordings
Experiments and Results
Conclusion
References
Non-negative Matrix Factorization Based Noise Reduction for Noise Robust Automatic Speech Recognition
Introduction
Proposed NMF-Based Noise Reduction Method for ASR
Stationary Noise Reduction Based on Wiener Filtering
Non-stationary Noise Reduction Based on NMF
Target Speech Reconstruction
Speech Recognition Experiments
Conclusion
References
Audio Imputation Using the Non-negative Hidden Markov Model
Introduction
Proposed Method
Overview
Probabilistic Model
Estimation of Incomplete Data
Experiments
Conclusions
References
A Non-negative Approach to Language Informed Speech Separation
Introduction
Models of Individual Speakers
Non-negative Hidden Markov Model
Word Models
Combining Word Models
Model of Mixtures
Combining Speaker Dependent Models
Speech Separation
Pruning
Experimental Results and Discussion
Conclusions
References
Temporally-Constrained Convolutive Probabilistic Latent Component Analysis for Multi-pitch Detection
Introduction
Proposed Method
Model
Parameter Estimation
Postprocessing
Evaluation
Datasets
Results
Conclusions
References
A Latently Constrained Mixture Model for Audio Source Separation and Localization
Introduction
Binaural Sound Representation
Constrained Mixtures for Separation and Localization
Experiments, Results, and Conclusions
References
Multiple Instrument Mixtures Source Separation Evaluation Using Instrument-Dependent NMF Models
Introduction
NMF Models
Basic Harmonic Constrained Model
Source-filter Model with Harmonic-Comb Excitation
Source-filter Model with Multi-Excitation Per Instrument
Application to Source Separation
Experimental Setup
Results
Conclusion and Perspectives
References
Complex Extension of Infinite Sparse Factor Analysis for Blind Speech Separation
Introduction
Blind Source Separation Using ISFA
Problem Settings of BSS
BSS for Speech Signals
Model of ISFA
Problems with Conventional Method
Complex Extension of ISFA
Priors
Likelihood Function
Posteriors
Postprocessing
Experimental Results
Conclusion
References
A General Framework for Online Audio Source Separation
Introduction
General Audio Source Separation Framework
Model
Offline EM-MU Algorithm
Online EM-MU Algorithm
Experimental Results
Conclusion
References
Sound Recognition in Mixtures
Introduction
Proposed Method
Basic Model
Modeling Temporal Dependencies
Experimental Results
Examples
Evaluation
Summary and Discussion
References
SiSEC 2011 Evaluation Campaign
The 2011 Signal Separation Evaluation Campaign (SiSEC2011): - Audio Source Separation -
Introduction
Specifications
Tasks
Datasets
Evaluation Criteria
Results
Conclusion
References
The 2011 Signal Separation Evaluation Campaign (SiSEC2011): - Biomedical Data Analysis -
Introduction
Estimating Causal Relations
Task
Dataset
Evaluation Criterion
Results
Cancer Pathway Reconstruction
The Task
Dataset
Evaluation
Results
Conclusion
References
Improved Perceptual Metrics for the Evaluation of Audio Source Separation
Introduction
The PEASS Metrics
Distortion Decomposition
PEMO-Q Component Saliences
Trained Nonlinear Mapping
Effect of the Design Parameters
Data and Evaluation Procedure
Main Results
Detailed Impact of the Mapping and the Version of PEMO-Q
Detailed Impact of the PEMO-Q Similarity Measure
Detailed Impact of the PEMO-Q Internal Representation
Conclusion and Perspectives
References
Musical Audio Source Separation Based on User-Selected F0 Track
Introduction
User-Guided Source Separation
Related Works
F0-Guided Musical Source Separation
Graphical User Interface
Ergonomy Issues
Practical Solutions
F0 Representation and Separation Algorithm
Audio Signal Model
F0 Line Selection and Usage
Separating the Selected Source
Experiments
Database and Protocoles
Usage Feedbacks
Separation Performance
Conclusion
References
A GMM Sound Source Model for Blind Speech Separation in Under-determined Conditions
Introduction
Under-determined Sound Source Separation
Problem Settings
Sound Source Model and Cost Function
Auxiliary Function Method
Derivation of Update Rules
Parameter Update Ordering
Experiments
Conclusion and Future Work
References
Model-Driven Speech Enhancement for Multisource Reverberant Environment (Signal Separation Evaluation Campaign (SiSEC) 2011)
Introduction
Problem Statement
Proposed Method
Noise Spectrum Estimation
ML Speech Estimation
Reconstructing the Separated Signals
Experiments and Results
Conclusion
References
Semi-blind Source Separation Based on ICA and Overlapped Speech Detection
Introduction
Problem Statement
Cancellation Filter
Building the CFB
Source Separation Using ICA and CFB
Separation by Adaptive Post-filtering
Experiments
Random Sources Activity in Unknown Static Positions
A Moving Source
Conclusion
References
Nonparametric Modelling of ECG: Applications to Denoising and to Single Sensor Fetal ECG Extraction
Introduction
Nonparametric Modelling of ECG
Denoising of ECG and Extraction of Fetal ECG from a Single Sensor
Numerical Experiments
Synthetic Data: ECG Denoising
Real Data: f-ECG Extraction
Conclusions and Perspectives
References
Other Applications
Nesterov's Iterations for NMF-Based Supervised Classification of Texture Patterns
Introduction
Image Preprocessing
NMF Algorithm
Nesterov's Iterations
Classification Results
Conclusions
References
Detection of Aliasing in Image Sequences Using Nonlinear Factor Analysis
Introduction
Nonlinear Factor Analysis for Modeling Image Offsets
Detection of Aliased Regions
Numerical Examples
References
Geometrical Method Using Simplicial Cones for Overdetermined Nonnegative Blind Source Separation: Application to Real PET Images
Introduction
Geometrical View of N-BSS Problem
Geometrical Method Using Simplicial Cones for Overdetermined Nonnegative Blind Source Separation
Determined Case : Simplicial Cone Shrinking Algorithm for Unmixing Nonnegative Sources (SCSA-UNS)
Proposed Method for Overdetermined Case
Simulations and Discussions
Conclusions an Future Works
References
Multi-domain Feature of Event-Related Potential Extracted by Nonnegative Tensor Factorization: 5 vs. 14 Electrodes EEG Data
Introduction
Method
Data Description
Nonnegative Tensor Factorization for Multi-domain Feature Extraction
Data Processing and Analysis
Results
Conclusions
References
The Use of Linear Feature Projection for Precipitation Classification Using Measurements from Commercial Microwave Links
Introduction
Method
Setup
Signal Processing
Experiment and Results
Discussion and Summary
References
Bayesian Inference of Latent Causes in Gene Regulatory Dynamics
Introduction
Bayesian Inference
Model Setup and Method Description
Data Description
Likelihood Setup
Prior Distributions
Results
Setup of the Toy Example
Results for the Toy Example
Comparison of Our Method with Standard Approaches
Conclusions and Outlook
References
Bayesian Fuzzy Clustering of Colored Graphs
Introduction
Method
Bayesian Fuzzy Clustering Model
Algorithm
Results
Algorithm Performance on a 2-Partite Toy Example
Fuzzy Clusters of a Protein-Complex Hypergraph
Conclusions
References
Author Index

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

Latent Variable Analysis and Signal Separation

Description

More details

Other editions

Additional editions

Content

System requirements