Feature learning and deep architectures

New directions for music informatics

Eric J. Humphrey, Juan P. Bello, Yann Lecun

Research output: Contribution to journalArticle

Abstract

As we look to advance the state of the art in content-based music informatics, there is a general sense that progress is decelerating throughout the field. On closer inspection, performance trajectories across several applications reveal that this is indeed the case, raising some difficult questions for the discipline: why are we slowing down, and what can we do about it Here, we strive to address both of these concerns. First, we critically review the standard approach to music signal analysis and identify three specific deficiencies to current methods: hand-crafted feature design is sub-optimal and unsustainable, the power of shallow architectures is fundamentally limited, and short-time analysis cannot encode musically meaningful structure. Acknowledging breakthroughs in other perceptual AI domains, we offer that deep learning holds the potential to overcome each of these obstacles. Through conceptual arguments for feature learning and deeper processing architectures, we demonstrate how deep processing models are more powerful extensions of current methods, and why now is the time for this paradigm shift. Finally, we conclude with a discussion of current challenges and the potential impact to further motivate an exploration of this promising research area.

Original languageEnglish (US)
Pages (from-to)461-481
Number of pages21
JournalJournal of Intelligent Information Systems
Volume41
Issue number3
DOIs
StatePublished - Dec 2013

Fingerprint

Signal analysis
Processing
Inspection
Trajectories
Deep learning

Keywords

  • Deep learning
  • Music informatics
  • Signal processing

ASJC Scopus subject areas

  • Artificial Intelligence
  • Information Systems
  • Hardware and Architecture
  • Computer Networks and Communications
  • Software

Cite this

Feature learning and deep architectures : New directions for music informatics. / Humphrey, Eric J.; Bello, Juan P.; Lecun, Yann.

In: Journal of Intelligent Information Systems, Vol. 41, No. 3, 12.2013, p. 461-481.

Research output: Contribution to journalArticle

@article{4ccc6c52c9f341758218f2a12ddefe49,
title = "Feature learning and deep architectures: New directions for music informatics",
abstract = "As we look to advance the state of the art in content-based music informatics, there is a general sense that progress is decelerating throughout the field. On closer inspection, performance trajectories across several applications reveal that this is indeed the case, raising some difficult questions for the discipline: why are we slowing down, and what can we do about it Here, we strive to address both of these concerns. First, we critically review the standard approach to music signal analysis and identify three specific deficiencies to current methods: hand-crafted feature design is sub-optimal and unsustainable, the power of shallow architectures is fundamentally limited, and short-time analysis cannot encode musically meaningful structure. Acknowledging breakthroughs in other perceptual AI domains, we offer that deep learning holds the potential to overcome each of these obstacles. Through conceptual arguments for feature learning and deeper processing architectures, we demonstrate how deep processing models are more powerful extensions of current methods, and why now is the time for this paradigm shift. Finally, we conclude with a discussion of current challenges and the potential impact to further motivate an exploration of this promising research area.",
keywords = "Deep learning, Music informatics, Signal processing",
author = "Humphrey, {Eric J.} and Bello, {Juan P.} and Yann Lecun",
year = "2013",
month = "12",
doi = "10.1007/s10844-013-0248-5",
language = "English (US)",
volume = "41",
pages = "461--481",
journal = "Journal of Intelligent Information Systems",
issn = "0925-9902",
publisher = "Springer Netherlands",
number = "3",

}

TY - JOUR

T1 - Feature learning and deep architectures

T2 - New directions for music informatics

AU - Humphrey, Eric J.

AU - Bello, Juan P.

AU - Lecun, Yann

PY - 2013/12

Y1 - 2013/12

N2 - As we look to advance the state of the art in content-based music informatics, there is a general sense that progress is decelerating throughout the field. On closer inspection, performance trajectories across several applications reveal that this is indeed the case, raising some difficult questions for the discipline: why are we slowing down, and what can we do about it Here, we strive to address both of these concerns. First, we critically review the standard approach to music signal analysis and identify three specific deficiencies to current methods: hand-crafted feature design is sub-optimal and unsustainable, the power of shallow architectures is fundamentally limited, and short-time analysis cannot encode musically meaningful structure. Acknowledging breakthroughs in other perceptual AI domains, we offer that deep learning holds the potential to overcome each of these obstacles. Through conceptual arguments for feature learning and deeper processing architectures, we demonstrate how deep processing models are more powerful extensions of current methods, and why now is the time for this paradigm shift. Finally, we conclude with a discussion of current challenges and the potential impact to further motivate an exploration of this promising research area.

AB - As we look to advance the state of the art in content-based music informatics, there is a general sense that progress is decelerating throughout the field. On closer inspection, performance trajectories across several applications reveal that this is indeed the case, raising some difficult questions for the discipline: why are we slowing down, and what can we do about it Here, we strive to address both of these concerns. First, we critically review the standard approach to music signal analysis and identify three specific deficiencies to current methods: hand-crafted feature design is sub-optimal and unsustainable, the power of shallow architectures is fundamentally limited, and short-time analysis cannot encode musically meaningful structure. Acknowledging breakthroughs in other perceptual AI domains, we offer that deep learning holds the potential to overcome each of these obstacles. Through conceptual arguments for feature learning and deeper processing architectures, we demonstrate how deep processing models are more powerful extensions of current methods, and why now is the time for this paradigm shift. Finally, we conclude with a discussion of current challenges and the potential impact to further motivate an exploration of this promising research area.

KW - Deep learning

KW - Music informatics

KW - Signal processing

UR - http://www.scopus.com/inward/record.url?scp=84888315556&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84888315556&partnerID=8YFLogxK

U2 - 10.1007/s10844-013-0248-5

DO - 10.1007/s10844-013-0248-5

M3 - Article

VL - 41

SP - 461

EP - 481

JO - Journal of Intelligent Information Systems

JF - Journal of Intelligent Information Systems

SN - 0925-9902

IS - 3

ER -