A novel objective function to optimize neural networks for emotion recognition from speech patterns

Kuan Chieh Huang, Yau-Hwang Kuo

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

Expressions of different emotions are usually overlapping and hard to distinguish. Besides, the amounts of feature patterns are usually imbalanced in the overlapping emotional expressions, and most conventional classifiers tend to prefer lager classes for archiving a better overall recognition rate. This drawback is also encountered in the Multi-layer perception (MLP) models frequently proposed for emotion recognition due to its superior classification capability and performance. However, MLP and most recognition techniques only refer to a mean square error and an overall error rate. Furthermore, using MLP has another disadvantage that needs to search a suitable network structure. In this paper, a novel objective function to optimize the MLP neural networks is proposed for solving these problems. This function considers the criteria of mean square error, classification error rate, and distances between the examples and the classification boundary, to optimize the network parameters and prune the links between neurons. Besides, the sigmoid and Gaussian transfer functions are adopted in our method to construct suitable classification boundaries. An artificial data set and the Danish emotional speech database are used to verify the MLP based classifier with the novel objective function. The experimental results show that the proposed model has better performance than conventional MLPs.

Original languageEnglish
Title of host publicationProceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010
Pages413-417
Number of pages5
DOIs
Publication statusPublished - 2010
Event2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010 - Kitakyushu, Japan
Duration: 2010 Dec 152010 Dec 17

Other

Other2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010
CountryJapan
CityKitakyushu
Period10-12-1510-12-17

Fingerprint

Emotion Recognition
Multilayer
Objective function
Optimise
Neural Networks
Neural networks
Mean square error
Classifiers
Overlapping
Error Rate
Beer
Classifier
Emotional Speech
Gaussian Function
Neurons
Transfer functions
Network Structure
Transfer Function
Neuron
Perception

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Theoretical Computer Science

Cite this

Huang, K. C., & Kuo, Y-H. (2010). A novel objective function to optimize neural networks for emotion recognition from speech patterns. In Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010 (pp. 413-417). [5716361] https://doi.org/10.1109/NABIC.2010.5716361
Huang, Kuan Chieh ; Kuo, Yau-Hwang. / A novel objective function to optimize neural networks for emotion recognition from speech patterns. Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010. 2010. pp. 413-417
@inproceedings{7d038eff5abb487c9c09f614d6e4f2b2,
title = "A novel objective function to optimize neural networks for emotion recognition from speech patterns",
abstract = "Expressions of different emotions are usually overlapping and hard to distinguish. Besides, the amounts of feature patterns are usually imbalanced in the overlapping emotional expressions, and most conventional classifiers tend to prefer lager classes for archiving a better overall recognition rate. This drawback is also encountered in the Multi-layer perception (MLP) models frequently proposed for emotion recognition due to its superior classification capability and performance. However, MLP and most recognition techniques only refer to a mean square error and an overall error rate. Furthermore, using MLP has another disadvantage that needs to search a suitable network structure. In this paper, a novel objective function to optimize the MLP neural networks is proposed for solving these problems. This function considers the criteria of mean square error, classification error rate, and distances between the examples and the classification boundary, to optimize the network parameters and prune the links between neurons. Besides, the sigmoid and Gaussian transfer functions are adopted in our method to construct suitable classification boundaries. An artificial data set and the Danish emotional speech database are used to verify the MLP based classifier with the novel objective function. The experimental results show that the proposed model has better performance than conventional MLPs.",
author = "Huang, {Kuan Chieh} and Yau-Hwang Kuo",
year = "2010",
doi = "10.1109/NABIC.2010.5716361",
language = "English",
isbn = "9781424473762",
pages = "413--417",
booktitle = "Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010",

}

Huang, KC & Kuo, Y-H 2010, A novel objective function to optimize neural networks for emotion recognition from speech patterns. in Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010., 5716361, pp. 413-417, 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010, Kitakyushu, Japan, 10-12-15. https://doi.org/10.1109/NABIC.2010.5716361

A novel objective function to optimize neural networks for emotion recognition from speech patterns. / Huang, Kuan Chieh; Kuo, Yau-Hwang.

Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010. 2010. p. 413-417 5716361.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A novel objective function to optimize neural networks for emotion recognition from speech patterns

AU - Huang, Kuan Chieh

AU - Kuo, Yau-Hwang

PY - 2010

Y1 - 2010

N2 - Expressions of different emotions are usually overlapping and hard to distinguish. Besides, the amounts of feature patterns are usually imbalanced in the overlapping emotional expressions, and most conventional classifiers tend to prefer lager classes for archiving a better overall recognition rate. This drawback is also encountered in the Multi-layer perception (MLP) models frequently proposed for emotion recognition due to its superior classification capability and performance. However, MLP and most recognition techniques only refer to a mean square error and an overall error rate. Furthermore, using MLP has another disadvantage that needs to search a suitable network structure. In this paper, a novel objective function to optimize the MLP neural networks is proposed for solving these problems. This function considers the criteria of mean square error, classification error rate, and distances between the examples and the classification boundary, to optimize the network parameters and prune the links between neurons. Besides, the sigmoid and Gaussian transfer functions are adopted in our method to construct suitable classification boundaries. An artificial data set and the Danish emotional speech database are used to verify the MLP based classifier with the novel objective function. The experimental results show that the proposed model has better performance than conventional MLPs.

AB - Expressions of different emotions are usually overlapping and hard to distinguish. Besides, the amounts of feature patterns are usually imbalanced in the overlapping emotional expressions, and most conventional classifiers tend to prefer lager classes for archiving a better overall recognition rate. This drawback is also encountered in the Multi-layer perception (MLP) models frequently proposed for emotion recognition due to its superior classification capability and performance. However, MLP and most recognition techniques only refer to a mean square error and an overall error rate. Furthermore, using MLP has another disadvantage that needs to search a suitable network structure. In this paper, a novel objective function to optimize the MLP neural networks is proposed for solving these problems. This function considers the criteria of mean square error, classification error rate, and distances between the examples and the classification boundary, to optimize the network parameters and prune the links between neurons. Besides, the sigmoid and Gaussian transfer functions are adopted in our method to construct suitable classification boundaries. An artificial data set and the Danish emotional speech database are used to verify the MLP based classifier with the novel objective function. The experimental results show that the proposed model has better performance than conventional MLPs.

UR - http://www.scopus.com/inward/record.url?scp=79952742688&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952742688&partnerID=8YFLogxK

U2 - 10.1109/NABIC.2010.5716361

DO - 10.1109/NABIC.2010.5716361

M3 - Conference contribution

SN - 9781424473762

SP - 413

EP - 417

BT - Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010

ER -

Huang KC, Kuo Y-H. A novel objective function to optimize neural networks for emotion recognition from speech patterns. In Proceedings - 2010 2nd World Congress on Nature and Biologically Inspired Computing, NaBIC 2010. 2010. p. 413-417. 5716361 https://doi.org/10.1109/NABIC.2010.5716361