Probabilisticmodelling of signal mixtures with differentiable dictionaries

Lukas Martak; Rainer Kelz; Gerhard Widmer

Probabilisticmodelling of signal mixtures with differentiable dictionaries

Lukas Martak, Rainer Kelz, Gerhard Widmer

Institute of Computational Perception

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

Abstract

We introduce a novel way to incorporate priorinformation into (semi-) supervised non-negative matrixfactorization, which we call differentiable dictionarysearch. It enables general, highly flexible and principledmodelling of mixtures where nonlinear sources are linearlymixed. We study its behavior on an audio decompositiontask, and conduct an extensive, highly controlled study ofits modelling capabilities

Original language	English
Title of host publication	Proceedings of the 29th European Signal Processing Conference(EUSIPCO 2021)
Number of pages	5
Publication status	Published - Aug 2021

Fields of science

202002 Audiovisual media
102 Computer Sciences
102001 Artificial intelligence
102003 Image processing
102015 Information systems

JKU Focus areas

Digital Transformation

Cite this

@inproceedings{a7104a9d91e546269f0e606bcadc2570,

title = "Probabilisticmodelling of signal mixtures with differentiable dictionaries",

abstract = "We introduce a novel way to incorporate priorinformation into (semi-) supervised non-negative matrixfactorization, which we call differentiable dictionarysearch. It enables general, highly flexible and principledmodelling of mixtures where nonlinear sources are linearlymixed. We study its behavior on an audio decompositiontask, and conduct an extensive, highly controlled study ofits modelling capabilities",

author = "Lukas Martak and Rainer Kelz and Gerhard Widmer",

year = "2021",

month = aug,

language = "English",

booktitle = "Proceedings of the 29th European Signal Processing Conference(EUSIPCO 2021)",

}

TY - GEN

T1 - Probabilisticmodelling of signal mixtures with differentiable dictionaries

AU - Martak, Lukas

AU - Kelz, Rainer

AU - Widmer, Gerhard

PY - 2021/8

Y1 - 2021/8

N2 - We introduce a novel way to incorporate priorinformation into (semi-) supervised non-negative matrixfactorization, which we call differentiable dictionarysearch. It enables general, highly flexible and principledmodelling of mixtures where nonlinear sources are linearlymixed. We study its behavior on an audio decompositiontask, and conduct an extensive, highly controlled study ofits modelling capabilities

AB - We introduce a novel way to incorporate priorinformation into (semi-) supervised non-negative matrixfactorization, which we call differentiable dictionarysearch. It enables general, highly flexible and principledmodelling of mixtures where nonlinear sources are linearlymixed. We study its behavior on an audio decompositiontask, and conduct an extensive, highly controlled study ofits modelling capabilities

M3 - Conference proceedings

BT - Proceedings of the 29th European Signal Processing Conference(EUSIPCO 2021)

ER -