• Home
  • Search
  • Monaural Speech Separation by Support Vector Machines: Bridging the Divide Between Supervised and Unsupervised Learning Methods
  • Cite Icon2
  • https://doi.org/10.1007/978-1-4020-6479-1_15Copy DOI Icon

Monaural Speech Separation by Support Vector Machines: Bridging the Divide Between Supervised and Unsupervised Learning Methods

  • Jan 1, 2007
  • Sepp Hochreiter +1 more
Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

We address the problem of identifying multiple independent speech sources from a single signal that is a mixture of the sources. Because the problem is ill-posed, standard independent component analysis (ICA) approaches which try to invert the mixing matrix fail. We show how the unsupervised problem can be transformed into a supervised regression task which is then solved by support- vector regression (SVR). It turns out that the linear SVR approach is equivalent to the sparse-decomposition method proposed by (1, 2). However, we can extend the method to nonlinear ICA by applying the trick. Beyond the kernel trick, the SVM perspective provides a new interpretation of the sparse-decomposition method's hyperparameter which is related to the input noise. The limitation of the SVM perspective is that, for the nonlinear case, it can recover only whether or not a mixture component is present; it cannot recover the strength of the component. In experiments, we show that our model can handle difficult problems and is especially well suited for speech signal separation.

Similar Papers
  • Research Article
  • Citations89

Independent component analysis applied to the removal of motion artifacts from electrocardiographic signals

  • Dec 07, 2007
  • Medical & Biological Engineering & Computing
  • M Milanesi +5
  • Research Article
  • Citations34

Independent Component Analysis–Based Feature Extraction Technique for Defect Classification Applied for Pulsed Eddy Current NDE

  • Oct 29, 2009
  • Research in Nondestructive Evaluation
  • Guang Yang +3
  • Research Article
  • Citations22

Erroneous Resting-State fMRI Connectivity Maps Due to Prolonged Arterial Arrival Time and How to Fix Them.

  • Aug 01, 2018
  • Brain Connectivity
  • Hesamoddin Jahanian +3
  • Research Article
  • Citations5

Dimensionality reduction of the attributes using fuzzy optimized independent component analysis for a Big Data Intrusion Detection System

  • Feb 01, 2022
  • Scientific and Technical Journal of Information Technologies, Mechanics and Optics
  • R Aswanandini +1
  • Research Article
  • Citations12

Support vector classification and regression for fault location in distribution system using voltage sag profile

  • Feb 03, 2017
  • IEEJ Transactions on Electrical and Electronic Engineering
  • Sophi Shilpa Gururajapathy +3
  • Research Article
  • Citations303

Reduced functional connectivity within and between ‘social’ resting state networks in autism spectrum conditions

  • Jun 08, 2012
  • Social Cognitive and Affective Neuroscience
  • Elisabeth A H Von Dem Hagen +3
  • PDF
  • Research Article
  • Citations16

Effects of Independent Component Analysis on Magnetoencephalography Source Localization in Pre-surgical Frontal Lobe Epilepsy Patients

  • Jun 02, 2020
  • Frontiers in Neurology
  • Giovanni Pellegrino +7
  • Research Article
  • Citations1

EMPIRICAL MODE DECOMPOSITION METHOD FOR MEG PHANTOM DATA ANALYSIS

  • Dec 01, 2009
  • Journal of Circuits, Systems and Computers
  • Juhong Yang +5
  • Conference Article
  • Citations3

Independent component analysis (ICA) and self-organizing map (SOM) approach to multidetection system for network intruders

  • Apr 04, 2003
  • Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE
  • Abdi M Abdi +1
  • Conference Article
  • Citations17

Coherency identification in interconnected power system - an independent component analysis approach

  • Jan 01, 2013
  • M A M Ariff +1
  • Research Article
  • Citations56

Characterization of the image-derived carotid artery input function using independent component analysis for the quantitation of [18F] fluorodeoxyglucose positron emission tomography images**Image-derived input function by ICA for FDG-PET.

  • Nov 15, 2007
  • Physics in Medicine & Biology
  • K Chen +6
  • Research Article

Classification of Sleep Apnea Types Using Clustering with SVM Classifier

  • Jan 01, 2012
  • International Journal of Biomedical and Clinical Engineering
  • Faiza Charfi +1
  • PDF
  • Research Article
  • Citations2

Evaluation of Dissolved Organic Carbon Using Synchronized Fluorescence Emission Spectra and Unsupervised Method of Principal Component Analysis (PCA) and Independent Component Analysis (ICA)

  • Jan 01, 2019
  • Journal of Water Resource and Protection
  • Tais Cristina Filippe +4
  • Research Article
  • Citations1

Independent Component Analysis for Image Recovery Using SOM-Based Noise Detection

  • Jun 01, 2007
  • IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
  • X Zhang +3
  • Research Article
  • Citations35

Time course based artifact identification for independent components of resting-state FMRI.

  • Jan 01, 2013
  • Frontiers in Human Neuroscience
  • Christian Rummel +6
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.