Abstract
The quantification of emotional states is an important step to understanding wellbeing. Time series data from multiple modalities such as physiological and motion sensor data have proven to be integral for measuring and quantifying emotions. Monitoring emotional trajectories over long periods of time inherits some critical limitations in relation to the size of the training data. This shortcoming may hinder the development of reliable and accurate machine learning models. To address this problem, this article proposes a framework to tackle the limitation in performing emotional state recognition: (1) encoding time series data into coloured images; (2) leveraging pre-trained object recognition models to apply a Transfer Learning (TL) approach using the images from step 1; (3) utilising a 1D Convolutional Neural Network (CNN) to perform emotion classification from physiological data; (4) concatenating the pre-trained TL model with the 1D CNN. We demonstrate that model performance when inferring real-world wellbeing rated on a 5-point Likert scale can be enhanced using our framework, resulting in up to 98.5% accuracy, outperforming a conventional CNN by 4.5%. Subject-independent models using the same approach resulted in an average of 72.3% accuracy (SD 0.038). The proposed methodology helps improve performance and overcome problems with small training datasets.
- [1] . 2023. Evaluating multimodal wearable sensors for quantifying affective states and depression with neural networks. IEEE Sensors Journal 23, 19 (2023), 22788–22802.
DOI: Google ScholarCross Ref - [2] . 2023. Evaluating different configurations of machine learning models and their transfer learning capabilities for stress detection using heart rate. Journal of Ambient Intelligence and Humanized Computing 14, 8 (2023), 11011–11021.
DOI: Google ScholarCross Ref - [3] . 2019. Toward user-independent emotion recognition using physiological signals. IEEE Sensors Journal 19, 19 (2019), 8402–8412.
DOI: Google ScholarCross Ref - [4] 2017. Emotion recognition based on EEG using LSTM recurrent neural network. International Journal of Advanced Computer Science and Applications 8, 10 (2017), 355–358.
DOI: Google ScholarCross Ref - [5] . 2012. A real-time model based Support Vector Machine for emotion recognition through EEG. In Proceedings of the 2012 International Conference on Control, Automation and Information Sciences, ICCAIS 2012.
DOI: Google ScholarCross Ref - [6] . 2018. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv:1803.01271. Retrieved from https://arxiv.org/abs/1803.01271Google Scholar
- [7] . 2017. A deep transfer learning approach for improved post-traumatic stress disorder diagnosis. In Proceedings of the IEEE International Conference on Data Mining, ICDM.Google ScholarCross Ref
- [8] . 2019. A deep transfer learning approach for improved post-traumatic stress disorder diagnosis. Knowledge and Information Systems 60, 3 (2019), 1693–1724.Google ScholarDigital Library
- [9] . 2014. A conceptual framework for integrated analysis of environmental quality and quality of life. Ecological Indicators 45 (2014), 664–668.
DOI: Google ScholarCross Ref - [10] . 2008. Releasing a thoroughly annotated and processed spontaneous emotional database: The FAU aibo emotion corpus. Proceedings of Workshop on Corpora for Research on Emotion and Affect LREC (2008), 28–31.Google Scholar
- [11] Thomas Boraud. 2020. How the Brain Makes Decisions (Oxford, 2020; online edn, Oxford Academic, 19 Nov. 2020). . Accessed 15 November 2023.Google ScholarCross Ref
- [12] . 2022. Predicting the internal knee abduction impulse during walking using deep learning. Frontiers in Bioengineering and Biotechnology 10 (2022), 1–9.
DOI: Google ScholarCross Ref - [13] . 2021. Improving the activity recognition using GMAF and transfer learning in post-stroke rehabilitation assessment. In Proceedings of the SAMI 2021—IEEE 19th World Symposium on Applied Machine Intelligence and Informatics.391–397.
DOI: Google ScholarCross Ref - [14] . 1994. Measuring emotion: The self-assessment manikin and the semantic differential. Journal of Behavior Therapy and Experimental Psychiatry 25, 1 (1994), 49–59.
DOI: Google ScholarCross Ref - [15] . 1997. Multitask learning. Machine Learning 28, 1 (1997), 41–75.
DOI: Google ScholarDigital Library - [16] . 2016. Detection of mental stress due to oral academic examination via ultra-short-term HRV analysis. In Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS.
DOI: Google ScholarCross Ref - [17] . 1998. Emotion in the perspective of an integrated nervous system. In Proceedings of the Brain Research Reviews.
DOI: Google ScholarCross Ref - [18] . 2020. Personalizing heart rate-based seizure detection using supervised SVM transfer learning. Frontiers in Neurology 11 (2020), 1–13. Google ScholarCross Ref
- [19] . 2010. Two stress detection schemes based on physiological signals for real-time applications. In Proceedings of the 2010 6th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIHMSP 2010.
DOI: Google ScholarDigital Library - [20] . 2017. Recognizing emotions from whispered speech based on acoustic feature transfer learning. IEEE Access 5 (2017), 1–1.
DOI: Google ScholarCross Ref - [21] . 2013. Sparse autoencoder-based feature transfer learning for speech emotion recognition. In Proceedings of the 2013 Humaine Association Conference on Affective Computing and Intelligent Interaction, ACII 2013. 511–516.
DOI: Google ScholarDigital Library - [22] . 2020. Image-based time series representations for pixelwise eucalyptus region classification: A comparative study. IEEE Geoscience and Remote Sensing Letters 17, 8 (2020), 1450–1454.
DOI: Google ScholarCross Ref - [23] . 1971. Constants across cultures in the face and emotion. Journal of Personality and Social Psychology 17, 2 (1971), 124–129.
DOI: Google ScholarCross Ref - [24] . 2012. DLPASCAL The pascal visual object classes challenge 2012 (VOC2012) development kit. Pattern Analysis, Statistical Modelling and Computational Learning, Tech. Rep (2012), 1–32.Google Scholar
- [25] . 2019. Inter-subject transfer learning with an end-to-end deep convolutional neural network for EEG-based BCI. Journal of Neural Engineering 16, 2 (2019), 1–12.
DOI: Google ScholarCross Ref - [26] . 2017. Deep Learning: Adaptive Computation and Machine Learning.Google Scholar
- [27] . 2019. 1D CNN-based transfer learning model for bearing fault diagnosis under variable working conditions. In Proceedings of the Advances in Intelligent Systems and Computing. Springer Verlag, 13–23.
DOI: Google ScholarCross Ref - [28] . 2005. Detecting stress during real-world driving tasks using physiological sensors. IEEE Transactions on Intelligent Transportation Systems 6, 2 (2005), 156–166.Google ScholarDigital Library
- [29] . 2015. Skin temperature reveals the intensity of acute stress. Physiology and Behavior 152, Pt A (2015), 225–230.
DOI: Google ScholarCross Ref - [30] . 2010. Social cognition and attitudes. Psychology (2010), 646–677.Google Scholar
- [31] . 2015. CStress: Towards a gold standard for continuous stress assessment in the mobile environment. In UbiComp 2015 - Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing.
DOI: Google ScholarDigital Library - [32] . 2019. Searching for MobileNetV3.
Technical Report .Google Scholar - [33] . 2012. Extreme learning machine for regression and multiclass classification. IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics 42, 2 (2012), 513–529.
DOI: Google ScholarDigital Library - [34] . 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine Learning, ICML 2015. International Machine Learning Society (IMLS), 448–456.Google ScholarDigital Library
- [35] . 2013. Real-time EEG-based happiness detection system. The Scientific World Journal 2013 (2013), 1–12.
DOI: Google ScholarCross Ref - [36] . 2018. EEG-based workers’ stress recognition at construction sites. Automation in Construction 93 (2018), 315–324.
DOI: Google ScholarCross Ref - [37] . 2018. Towards unravelling the relationship between on-body, environmental and emotion data using sensor information fusion approach. Information Fusion 40 (2018), 18–31.Google ScholarCross Ref
- [38] . 2019. Deep learning analysis of mobile physiological, environmental and location sensor data for emotion detection. Information Fusion 49 (2019), 46–56.Google ScholarDigital Library
- [39] . 2019. Scalable classification of univariate and multivariate time series. In Proceedings of the 2018 IEEE International Conference on Big Data, Big Data 2018.
DOI: Google ScholarCross Ref - [40] . 2011. A brain-computer interface for classifying EEG correlates of chronic mental stress. In Proceedings of the 2011 International Joint Conference on Neural Networks. IEEE, 757–762.
DOI: Google ScholarCross Ref - [41] . 2016. Utilizing a wristband sensor to measure the stress level for people with dementia. Sensors (Basel, Switzerland) 16, 12 (2016), 1–17.
DOI: Google ScholarCross Ref - [42] . 2019. Convolutional neural network coupled with a transfer-learning approach for time-series flood predictions. Water 12, 1 (2019), 96.Google ScholarCross Ref
- [43] . 2019. A pilot randomized trial of five financial incentive strategies to increase study enrollment and retention rates. Contemporary Clinical Trials Communications 15 (2019), 1–5.
DOI: Google ScholarCross Ref - [44] . 2011. Activity recognition using cell phone accelerometers. ACM SIGKDD Explorations Newsletter 12, 2 (2011), 74–82.
DOI: Google ScholarDigital Library - [45] . 2019. Cross-subject driver status detection from physiological signals based on hybrid feature selection and transfer learning. Expert Systems with Applications 137 (2019), 266–280.
DOI: Google ScholarDigital Library - [46] . 2018. Transfer learning for improving speech emotion classification accuracy. In Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH.
DOI: Google ScholarCross Ref - [47] . 2019. Multisource transfer learning for cross-subject eeg emotion recognition. IEEE Transactions on Cybernetics 50, 7 (2019), 3281–3293.Google Scholar
- [48] . 2017. Improving EEG-based emotion classification using conditional transfer learning. Frontiers in Human Neuroscience 11 (2017), 1–11.Google Scholar
- [49] . 2010. EEG-based emotion recognition in music listening. IEEE Transactions on Biomedical Engineering (2010).
DOI: Google ScholarCross Ref - [50] . 2018. Psychological stress level detection based on electrodermal activity. Behavioural Brain Research 341 (2018), 50–53.Google ScholarCross Ref
- [51] . 2011. Researching emotion: Challenges and solutions. In Proceedings of the 2011 iConference on - iConference ’11. ACM, New York, New York.
DOI: Google ScholarDigital Library - [52] . 2013. Learning deep physiological models of affect. IEEE Computational Intelligence Magazine 8, 2 (2013), 20–33.Google ScholarDigital Library
- [53] . 2016. Stress modelling and prediction in presence of scarce data. Journal of Biomedical Informatics 63 (2016), 344–356.Google ScholarDigital Library
- [54] . 2019. Microsoft Band 2 Features and Functions. Retrieved from https://support.microsoft.com/en-gb/help/4000313. Accessed 16th November 2023.Google Scholar
- [55] . 2022. A deep transfer learning approach to an effective classification of water pollutants from voltammetric characterizations. In Proceedings of the MELECON 2022—IEEE Mediterranean Electrotechnical Conference.255–259.
DOI: Google ScholarCross Ref - [56] . 1960. Learning Theory and Behavior.John Wiley and Sons Inc.
DOI: Google ScholarCross Ref - [57] . 2013. Towards measuring stress with smartphones and wearable devices during workday and sleep. BioNanoScience 3, 2 (2013), 172–183.
DOI: Google ScholarCross Ref - [58] . 2015. Deep learning for emotion recognition on small datasets using transfer learning. In ICMI 2015—Proceedings of the 2015 ACM International Conference on Multimodal Interaction. Association for Computing Machinery, Inc, 443–449.
DOI: Google ScholarDigital Library - [59] . 2017. Audio-visual emotion recognition using deep transfer learning and multiple temporal models. In ICMI 2017—Proceedings of the 19th ACM International Conference on Multimodal Interaction. Association for Computing Machinery, Inc, 577–582.
DOI: Google ScholarDigital Library - [60] Sinno Jialin Pan and Yang Qiang. 2009. A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering 22, 10 (2009), 1345–1359.Google Scholar
- [61] . 2022. Hybrid deep convolutional model-based emotion recognition using multiple physiological signals. Computer Methods in Biomechanics and Biomedical Engineering 25, 15 (2022), 1678–1690.
DOI: Google ScholarCross Ref - [62] . 2018. Experiencing mental health diagnosis: A systematic review of service user, clinician, and carer perspectives across clinical settings. The Lancet Psychiatry 5, 9 (2018), 747–764.Google Scholar
- [63] . 2003. Affective computing: Challenges. International Journal of Human-Computer Studies 59, 1–2 (2003), 55–64.
DOI: Google ScholarDigital Library - [64] . 2017. A novel deep-learning based framework for multi-subject emotion recognition. In Proceedings of the ICCSS 2017—2017 International Conference on Information, Cybernetics, and Computational Social Systems. IEEE, 181–185.Google ScholarCross Ref
- [65] . 2018. Emotion recognition based on gramian encoding visualization. In Proceedings of the Brain Informatics: International Conference, BI 2018. 3–12.Google Scholar
- [66] . 2020. Transfer learning for subject-independent stress detection using physiological signals. In Proceedings of the 2020 IEEE 17th India Council International Conference, INDICON 2020.
DOI: Google ScholarCross Ref - [67] . 1980. A circumplex model of affect. Journal of Personality and Social Psychology 39, 6 (1980), 1161–1178.
DOI: Google ScholarCross Ref - [68] . 2015. Multimodal affect databases: Collection, challenges, and chances. Handbook of Affective Computing (2015), 323–333.Google Scholar
- [69] . 2010. Discriminating stress from cognitive load using a wearable EDA device. IEEE Transactions on Information Technology in Biomedicine 14, 2 (2010), 410–417.Google Scholar
- [70] . 2017. Transforming sensor data to the image domain for deep learning-an application to footstep detection. arXiv:1701.01077v3. Retrieved from https://arxiv.org/abs/1701.01077v3Google Scholar
- [71] . 2012. Objective measures, sensors and computational techniques for stress recognition and classification: A survey. Computer Methods and Programs in Biomedicine 108, 3 (2012), 1287–1301.Google ScholarDigital Library
- [72] . 2017. Hashtag healthcare: from tweets to mental health journals using deep transfer learning. arXiv preprint arXiv:1708.01372 (2017).Google Scholar
- [73] . 2015. Very deep convolutional networks for large-scale image recognition. In Proceedings of the 3rd International Conference on Learning Representations, ICLR 2015.Google Scholar
- [74] . 2014. Dropout: A Simple Way to Prevent Neural Networks from Overfitting.
Technical Report .Google Scholar - [75] . 2016. Rethinking the inception architecture for computer vision. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition.
DOI: Google ScholarCross Ref - [76] . 2008. Influence of mental stress on heart rate and heart rate variability. In Proceedings of the IFMBE.1366–1369.
DOI: Google ScholarCross Ref - [77] . 2021. Coal-rock interface recognition method based on GAF-deep learning. In Proceedings of the 5th IEEE Conference on Energy Internet and Energy System Integration: Energy Internet for Carbon Neutrality, EI2 2021.4029–4033.
DOI: Google ScholarCross Ref - [78] . 2021. Drought prediction based on feature-based transfer learning and time series imaging. IEEE Access 9 (2021), 101454–101468.
DOI: Google ScholarCross Ref - [79] . 2010. Safety in numbers: Learning categories from few examples with multi model knowledge transfer. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition.
DOI: Google ScholarCross Ref - [80] . 2013. A methodology for the analysis of medical data. Handbook of Systems and Complexity in Health (2013), 113–125.
DOI: Google ScholarCross Ref - [81] . 2016. Daily longitudinal self-monitoring of mood variability in bipolar disorder and borderline personality disorder. Journal of Affective Disorders 205 (2016), 225–233.Google ScholarCross Ref
- [82] . 2019. Improving students’ daily life stress forecasting using LSTM neural networks. In 2019 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI). IEEE, 1–4.Google Scholar
- [83] . 2020. Signal to image to classification: Transfer learning for ECG. In Proceedings of the 2020 11th Conference of the European Study Group on Cardiovascular Oscillations: Computation and Modelling in Physiology: New Challenges and Opportunities, ESGCO 2020.
DOI: Google ScholarCross Ref - [84] . 2018. Deep transfer learning for cross-domain activity recognition. In Proceedings of the 3rd International Conference on Crowd Science and Engineering.
DOI: arxiv:1807.07963 Google ScholarDigital Library - [85] . 2022. Cuff-less blood pressure estimation from photoplethysmography via visibility graph and transfer learning. IEEE Journal of Biomedical and Health Informatics 26, 5 (2022), 2075–2085.
DOI: Google ScholarCross Ref - [86] . 2015. Imaging time-series to improve classification and imputation. In Proceedings of the IJCAI International Joint Conference on Artificial Intelligence.Google Scholar
- [87] . 1985. An attributional approach to emotional development. In Proceedings of the Emotions, Cognition, and Behavior.Google Scholar
- [88] . 2014. Emotion recognition based on multi-variant correlation of physiological signals. IEEE Transactions on Affective Computing 5, 2 (2014), 126–140.
DOI: Google ScholarCross Ref - [89] . 2021. Process outcome prediction: CNN vs. LSTM (with attention). Lecture Notes in Business Information Processing 397 (2021), 321–333.
DOI: Google ScholarCross Ref - [90] . 2014. WHO | Mental Health: A State of Well-Being. Retrieved from https://www.who.int/features/factfiles/mental_health/en/. Accessed 16th November 2023.Google Scholar
- [91] . 2011. Towards mental stress detection using wearable physiological sensors. In Proceedings of the 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE, 1798–1801.Google ScholarCross Ref
- [92] . 2020. Beyond mobile apps: A survey of technologies for mental well-being. IEEE Transactions on Affective Computing 13, 3 (2020), 1216–1235.Google ScholarCross Ref
- [93] . 2021. Towards personalised mental wellbeing recognition on-device using transfer learning “in the wild”. In Proceedings of the IEEE International Smart Cities Conference 2021.Google Scholar
- [94] . 2020. LabelSens: Enabling real-time sensor data labelling at the point of collection using an artificial intelligence-based approach. Personal and Ubiquitous Computing 24, 5 (2020), 709–722.Google ScholarDigital Library
- [95] . 2019. SAE+LSTM: A new framework for emotion recognition from multi-channel EEG. Frontiers in Neurorobotics 13 (2019), 1–14.Google ScholarCross Ref
- [96] . 2019. Multivariate time series data transformation for convolutional neural network. In Proceedings of the 2019 IEEE/SICE International Symposium on System Integration, SII 2019. Institute of Electrical and Electronics Engineers Inc., 188–192.
DOI: Google ScholarCross Ref - [97] . 2013. Performance and challenges in utilizing non-intrusive sensors for traffic data collection. Advances in Remote Sensing 2, 2 (2013), 45–50.
DOI: Google ScholarCross Ref - [98] . 2016. HealthyOffice: Mood recognition at work using smartphones and wearable sensors. In Proceedings of the 2016 IEEE International Conference on Pervasive Computing and Communication Workshops, PerCom Workshops 2016.
DOI: Google ScholarCross Ref - [99] . 2023. Global spatial representation: EEG correcting for subject-independent emotion recognition. In Proceedings of the International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems. Springer, Cham, 385–396.
DOI: Google ScholarDigital Library - [100] . 2017. Transfer learning with neural networks for bearing fault diagnosis in changing working conditions. IEEE Access 5 (2017), 14347–14357.Google ScholarCross Ref
- [101] . 2020. An investigation of deep learning models for eeg-based emotion recognition. Frontiers in Neuroscience 14 (2020), 1344.
DOI: Google ScholarCross Ref - [102] . 2016. Personalizing EEG-based affective models with transfer learning. In Proceedings of the 25th International Joint Conference on Artificial Intelligence (IJCAI-16).Google Scholar
- [103] . 2014. Privacy in the internet of things: Threats and challenges. Security and Communication Networks 7 (2014), 2728–2742.
DOI: Google ScholarCross Ref - [104] . 2023. LSTM-modeling of emotion recognition using peripheral physiological signals in naturalistic conversations. IEEE Journal of Biomedical and Health Informatics 27, 2 (2023), 912–923.
DOI: Google ScholarCross Ref
Index Terms
- Combining Deep Learning with Signal-image Encoding for Multi-Modal Mental Wellbeing Classification
Recommendations
Analysis of facial emotion expression in eating occasions using deep learning
AbstractEating is experienced as an emotional social activity in any culture. There are factors that influence the emotions felt during food consumption. The emotion felt while eating has a significant impact on our lives and affects different health ...
Group emotion recognition in the wild by combining deep neural networks for facial expression classification and scene-context analysis
ICMI '17: Proceedings of the 19th ACM International Conference on Multimodal InteractionThis paper presents the implementation details of a proposed solution to the Emotion Recognition in the Wild 2017 Challenge, in the category of group-level emotion recognition. The objective of this sub-challenge is to classify a group's emotion as ...
Biometric valence and arousal recognition
OZCHI '07: Proceedings of the 19th Australasian conference on Computer-Human Interaction: Entertaining User InterfacesA real-time user-independent emotion detection system using physiological signals has been developed. The system has the ability to classify affective states into 2-dimensions using valence and arousal. Each dimension ranges from 1 to 5 giving a total ...
Comments