Self-structuring hidden control neural models

Helge Bjarup Dissing Sørensen; U. Hartmann

doi:10.1109/NNSP.1992.253698

Self-structuring hidden control neural models

Helge Bjarup Dissing Sørensen
, U. Hartmann

Aalborg University

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

Abstract

The majority of neural models for pattern recognition have fixed architecture during training. A typical consequence is non-optimal and often too large networks. In this paper we propose a Self-structuring Hidden Control (SHC) neural model for pattern recognition, which establishes a near optimal architecture during training. We typically achieve a significant network architecture reduction in terms of the number of hidden Processing Elements (PE). The SHC model combines self-structuring architecture generation with non-linear prediction and hidden Markov modelling. The paper presents a theorem for self-structuring neural models stating that these models are universal approximators and thus relevant for real-world pattern recognition. Using SHC models containing as few as five hidden PES each for an isolated word recognition task resulted in a recognition rate of 98.4%. SHC models can furthermore be applied to continuous speech recognition.

Original language	English
Title of host publication	Neural Networks for Signal Processing II : Proceedings of the 1992 IEEE-SP Workshop
Publisher	IEEE Press
Publication date	1992
Pages	149-156
ISBN (Print)	0-7803-0557-4
DOIs	https://doi.org/10.1109/NNSP.1992.253698
Publication status	Published - 1992
Externally published	Yes
Event	1992 IEEE Workshop on Neural Networks for Signal Processing - Hotel Marielyst, Helsingoer, Denmark Duration: 31 Aug 1992 → 2 Sept 1992 http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=631

Conference

Conference	1992 IEEE Workshop on Neural Networks for Signal Processing
Location	Hotel Marielyst
Country/Territory	Denmark
City	Helsingoer
Period	31/08/1992 → 02/09/1992
Internet address	http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=631

Access to Document

10.1109/NNSP.1992.253698

OpenUrl availability

Check availability in DTU Findit

Cite this

@inproceedings{4a11548e418b40cbaabcc7ca5a5f9aa7,

title = "Self-structuring hidden control neural models",

abstract = "The majority of neural models for pattern recognition have fixed architecture during training. A typical consequence is non-optimal and often too large networks. In this paper we propose a Self-structuring Hidden Control (SHC) neural model for pattern recognition, which establishes a near optimal architecture during training. We typically achieve a significant network architecture reduction in terms of the number of hidden Processing Elements (PE). The SHC model combines self-structuring architecture generation with non-linear prediction and hidden Markov modelling. The paper presents a theorem for self-structuring neural models stating that these models are universal approximators and thus relevant for real-world pattern recognition. Using SHC models containing as few as five hidden PES each for an isolated word recognition task resulted in a recognition rate of 98.4\%. SHC models can furthermore be applied to continuous speech recognition.",

author = "S{\o}rensen, \{Helge Bjarup Dissing\} and U. Hartmann",

year = "1992",

doi = "10.1109/NNSP.1992.253698",

language = "English",

isbn = "0-7803-0557-4",

pages = "149--156",

booktitle = "Neural Networks for Signal Processing II",

publisher = "IEEE Press",

note = "1992 IEEE Workshop on Neural Networks for Signal Processing, IEEE NNSP 1992 ; Conference date: 31-08-1992 Through 02-09-1992",

url = "http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=631",

}

TY - GEN

T1 - Self-structuring hidden control neural models

AU - Sørensen, Helge Bjarup Dissing

AU - Hartmann, U.

PY - 1992

Y1 - 1992

N2 - The majority of neural models for pattern recognition have fixed architecture during training. A typical consequence is non-optimal and often too large networks. In this paper we propose a Self-structuring Hidden Control (SHC) neural model for pattern recognition, which establishes a near optimal architecture during training. We typically achieve a significant network architecture reduction in terms of the number of hidden Processing Elements (PE). The SHC model combines self-structuring architecture generation with non-linear prediction and hidden Markov modelling. The paper presents a theorem for self-structuring neural models stating that these models are universal approximators and thus relevant for real-world pattern recognition. Using SHC models containing as few as five hidden PES each for an isolated word recognition task resulted in a recognition rate of 98.4%. SHC models can furthermore be applied to continuous speech recognition.

AB - The majority of neural models for pattern recognition have fixed architecture during training. A typical consequence is non-optimal and often too large networks. In this paper we propose a Self-structuring Hidden Control (SHC) neural model for pattern recognition, which establishes a near optimal architecture during training. We typically achieve a significant network architecture reduction in terms of the number of hidden Processing Elements (PE). The SHC model combines self-structuring architecture generation with non-linear prediction and hidden Markov modelling. The paper presents a theorem for self-structuring neural models stating that these models are universal approximators and thus relevant for real-world pattern recognition. Using SHC models containing as few as five hidden PES each for an isolated word recognition task resulted in a recognition rate of 98.4%. SHC models can furthermore be applied to continuous speech recognition.

U2 - 10.1109/NNSP.1992.253698

DO - 10.1109/NNSP.1992.253698

M3 - Article in proceedings

SN - 0-7803-0557-4

SP - 149

EP - 156

BT - Neural Networks for Signal Processing II

PB - IEEE Press

T2 - 1992 IEEE Workshop on Neural Networks for Signal Processing

Y2 - 31 August 1992 through 2 September 1992

ER -

Self-structuring hidden control neural models

Abstract

Conference

Access to Document

OpenUrl availability

Fingerprint

Cite this