AI-Powered Emotional Recognition Systems: Applications and Limits

  • Authors

    • Ajay Krishnan Technical Lead, Tech Mahindra, India. Author

    DOI:

    https://doi.org/10.67228/3071561X/IJIRHT-2019PII9W1F

    Published 10-03-2019

  • Affective Computing, Emotion Recognition, Artificial Intelligence, Deep Learning, Facial Expression Analysis, Multimodal Systems, Human–Computer Interaction

    Issue

    Vol. 2 No. 2 (2019)

    Section

    Articles

    How to Cite

    [1]
    A. Krishnan, “AI-Powered Emotional Recognition Systems: Applications and Limits”, IJIRHT, vol. 2, no. 2, pp. 01–13, Oct. 2019, doi: 10.67228/3071561X/IJIRHT-2019PII9W1F.

  • Abstract

    Machine learning, affective computing, computer vision, and signal processing have become key important research fields due to the development of Artificial Intelligence (AI)-based emotional recognition systems. The purpose of these systems is automatic detection and decoding of the human emotional conditions based on the evidence of facial expression, speech, physiological activity, and textual expressions. The trend of installation of emotion-aware systems in healthcare, education, human-computer interaction (HCI), and security applications demonstrates their rise in the importance of society. But even with impressive technological breakthroughs, there are still significant issues concerning reliability, data biasing, interpretability, privacy and ethical implementation. The study outlined in this paper is a full-fledged study of the AI-based emotional recognition systems by examining their theoretical framework, areas of application, methodological frameworks, and the limitations of such systems. An organized literature review is described to follow how emotion recognition technologies have evolved since then and pinpoint dominating models and datasets. The approach suggested combines multimodal feature extraction of the data with deep learning classifiers to increase the level of the recognition. Performance under different conditions of a system is evaluated through experimental evaluation using benchmark datasets. Findings indicate that deep learning architectures are highly classified but the performance of the 18 system decreases considerably under real-life uncontrollable conditions.  Moreover, in this paper, the limitations inherent in the emotional recognition systems will be described such as cultural dependence, emotional subjectivity, imbalance of a dataset, and such issues as misuse of surveillance and emotional profiling will be raised as ethical considerations. The conclusion of the paper underlines the need to have open, elucidable, and ethics-controlled AI solutions to emotion-mindful techs. The proposed avenues of future research include strong multimodal learning, domain adaptation, and responsible AI governance. The results add to the knowledge regarding the practical and theoretical limitations of the AI-based emotional recognition systems.

  • References

    [1] Ekman, P. (1992). An argument for basic emotions. Cognition & Emotion, 6(3–4), 169–200.

    [2] Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39(6), 1161–1178.

    [3] Poria, S., Cambria, E., Bajpai, R., & Hussain, A. (2017). A review of affective computing: From unimodal analysis to multimodal fusion. Information Fusion, 37, 98–125.

    [4] Zeng, Z., Pantic, M., Roisman, G. I., & Huang, T. S. (2009). A survey of affect recognition methods: Audio, visual, and spontaneous expressions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 31(1), 39–58.

    [5] Shan, C., Gong, S., & McOwan, P. W. (2009). Facial expression recognition based on Local Binary Patterns: A comprehensive study. Image and Vision Computing, 27(6), 803–816.

    [6] Schuller, B., Steidl, S., Batliner, A., et al. (2009). The INTERSPEECH 2009 emotion challenge. Proceedings of Interspeech, 312–315.

    [7] Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine Learning, 20(3), 273–297.

    [8] LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444.

    [9] Tang, Y. (2013). Deep learning using linear support vector machines. IEEE Workshop on Challenges in Representation Learning.

    [10] Goodfellow, I., et al. (2013). Challenges in representation learning: A report on three machine learning contests. Neural Networks, 64, 59–63. (FER-2013 dataset)

    [11] Lucey, P., et al. (2010). The Extended Cohn-Kanade Dataset (CK+): A complete dataset for action unit and emotion-specified expression. IEEE CVPR Workshops, 94–101.

    [12] Busso, C., et al. (2008). IEMOCAP: Interactive emotional dyadic motion capture database. Language Resources and Evaluation, 42(4), 335–359.

    [13] Koelstra, S., et al. (2012). DEAP: A database for emotion analysis using physiological signals. IEEE Transactions on Affective Computing, 3(1), 18–31.

    [14] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780.

    [15] Mehrabian, A. (1996). Pleasure-arousal-dominance: A general framework for describing and measuring individual differences in temperament. Current Psychology, 14(4), 261–292.

  • Downloads