Using feature-level fusion for multimodal gender recognition for opinion mining videos

Sadam Al-Azani; El Sayed M. El-Alfy

doi:10.1109/ICMSAO.2019.8902177

Using feature-level fusion for multimodal gender recognition for opinion mining videos

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

1 Scopus citations

Abstract

Video has rich content and has recently received considerable attention for opinion mining and sentiment analysis, especially with the growing volume of online videos in social networks. However, it has been proved that machine-learning techniques for sentiment analysis are gender biased, i.e. can be more accurate for one gender than the other. Hence, if gender can be recognized early, it can lead to improving the sentiment analysis results. This paper explores multimodal analysis of video data to extract features and build gender recognition models. Different methods are evaluated for unimodal, bimodal and trimodal systems using features from three modalities: visual, audio, and text. The proposed system is evaluated and the results revealed that the fusion of multiple modalities at the feature level can have significant impact on the performance of gender recognition as compared to unimodal models.

Original language	English
Title of host publication	2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781538676844
DOIs	https://doi.org/10.1109/ICMSAO.2019.8902177
State	Published - Apr 2019

Publication series

Name	2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019
Volume	2019-January

Bibliographical note

Publisher Copyright:
© 2019 IEEE.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 5 Gender Equality

ASJC Scopus subject areas

Signal Processing
Industrial and Manufacturing Engineering
Safety, Risk, Reliability and Quality
Control and Optimization
Modeling and Simulation
Health Informatics

Access to Document

10.1109/ICMSAO.2019.8902177

Cite this

Al-Azani, S., & El-Alfy, E. S. M. (2019). Using feature-level fusion for multimodal gender recognition for opinion mining videos. In 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019 Article 8902177 (2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019; Vol. 2019-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICMSAO.2019.8902177

Al-Azani, Sadam ; El-Alfy, El Sayed M. / Using feature-level fusion for multimodal gender recognition for opinion mining videos. 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019. Institute of Electrical and Electronics Engineers Inc., 2019. (2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019).

@inproceedings{a44d6969fd5b42518cfa9414ed569769,

title = "Using feature-level fusion for multimodal gender recognition for opinion mining videos",

abstract = "Video has rich content and has recently received considerable attention for opinion mining and sentiment analysis, especially with the growing volume of online videos in social networks. However, it has been proved that machine-learning techniques for sentiment analysis are gender biased, i.e. can be more accurate for one gender than the other. Hence, if gender can be recognized early, it can lead to improving the sentiment analysis results. This paper explores multimodal analysis of video data to extract features and build gender recognition models. Different methods are evaluated for unimodal, bimodal and trimodal systems using features from three modalities: visual, audio, and text. The proposed system is evaluated and the results revealed that the fusion of multiple modalities at the feature level can have significant impact on the performance of gender recognition as compared to unimodal models.",

author = "Sadam Al-Azani and El-Alfy, \{El Sayed M.\}",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.",

year = "2019",

month = apr,

doi = "10.1109/ICMSAO.2019.8902177",

language = "English",

series = "2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019",

address = "United States",

}

Al-Azani, S & El-Alfy, ESM 2019, Using feature-level fusion for multimodal gender recognition for opinion mining videos. in 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019., 8902177, 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019, vol. 2019-January, Institute of Electrical and Electronics Engineers Inc. https://doi.org/10.1109/ICMSAO.2019.8902177

Using feature-level fusion for multimodal gender recognition for opinion mining videos. / Al-Azani, Sadam ; El-Alfy, El Sayed M.
2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019. Institute of Electrical and Electronics Engineers Inc., 2019. 8902177 (2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019; Vol. 2019-January).

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

TY - GEN

T1 - Using feature-level fusion for multimodal gender recognition for opinion mining videos

AU - Al-Azani, Sadam

AU - El-Alfy, El Sayed M.

PY - 2019/4

Y1 - 2019/4

N2 - Video has rich content and has recently received considerable attention for opinion mining and sentiment analysis, especially with the growing volume of online videos in social networks. However, it has been proved that machine-learning techniques for sentiment analysis are gender biased, i.e. can be more accurate for one gender than the other. Hence, if gender can be recognized early, it can lead to improving the sentiment analysis results. This paper explores multimodal analysis of video data to extract features and build gender recognition models. Different methods are evaluated for unimodal, bimodal and trimodal systems using features from three modalities: visual, audio, and text. The proposed system is evaluated and the results revealed that the fusion of multiple modalities at the feature level can have significant impact on the performance of gender recognition as compared to unimodal models.

AB - Video has rich content and has recently received considerable attention for opinion mining and sentiment analysis, especially with the growing volume of online videos in social networks. However, it has been proved that machine-learning techniques for sentiment analysis are gender biased, i.e. can be more accurate for one gender than the other. Hence, if gender can be recognized early, it can lead to improving the sentiment analysis results. This paper explores multimodal analysis of video data to extract features and build gender recognition models. Different methods are evaluated for unimodal, bimodal and trimodal systems using features from three modalities: visual, audio, and text. The proposed system is evaluated and the results revealed that the fusion of multiple modalities at the feature level can have significant impact on the performance of gender recognition as compared to unimodal models.

UR - https://www.scopus.com/pages/publications/85075783145

U2 - 10.1109/ICMSAO.2019.8902177

DO - 10.1109/ICMSAO.2019.8902177

M3 - Conference contribution

AN - SCOPUS:85075783145

T3 - 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019

BT - 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Al-Azani S , El-Alfy ESM. Using feature-level fusion for multimodal gender recognition for opinion mining videos. In 2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019. Institute of Electrical and Electronics Engineers Inc. 2019. 8902177. (2019 8th International Conference on Modeling Simulation and Applied Optimization, ICMSAO 2019). doi: 10.1109/ICMSAO.2019.8902177

Using feature-level fusion for multimodal gender recognition for opinion mining videos

Abstract

Publication series

Bibliographical note

UN SDGs

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this