Hierarchical Autoencoder Frequency Features for Stress Detection
Stress has a significant negative impact on people, which has made it a primary social concern. Early stress detection is essential for effective stress management. This study proposes a Deep Learning (DL) method for effective stress detection using multimodal physiological signals - Electrocardiogr...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10254202/ |
| _version_ | 1797668066465677312 |
|---|---|
| author | Radhika Kuttala Ramanathan Subramanian Venkata Ramana Murthy Oruganti |
| author_facet | Radhika Kuttala Ramanathan Subramanian Venkata Ramana Murthy Oruganti |
| author_sort | Radhika Kuttala |
| collection | DOAJ |
| description | Stress has a significant negative impact on people, which has made it a primary social concern. Early stress detection is essential for effective stress management. This study proposes a Deep Learning (DL) method for effective stress detection using multimodal physiological signals - Electrocardiogram (ECG) and Electrodermal activity (EDA). The extensive latent feature representation of DL models has yet to be fully explored. Hence, this paper proposes a hierarchical AutoEncoder (AE) feature fusion on the frequency domain. The latent representations from different layers of the autoencoder are combined and given as input to the classifier - Convolutional Recurrent Neural Network with Squeeze and Excitation (CRNN-SE) model. A two-set performance comparison is performed (romannum 1) performance on frequency band features, and raw data are compared. (romannum 2) autoencoders trained on three cost functions - Mean Squared Error (MSE), Kullback-Leibler (KL) divergence, and Cosine similarity performance are compared on frequency band features and raw data. To verify the generalizability of our approach, we tested it on four benchmark datasets- WAUC, CLAS, MAUS and ASCERTAIN. Results show that frequency band features showed better results than raw data by 4-8%, respectively. MSE loss produced better results than other losses for both frequency band features and raw data by 3-7%, respectively. The proposed approach considerably outperforms existing stress detection models that are subject-independent by 1-2%, respectively. |
| first_indexed | 2024-03-11T20:22:40Z |
| format | Article |
| id | doaj.art-ca7fc297d52c4e2e9c8e139a3de715b9 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-11T20:22:40Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-ca7fc297d52c4e2e9c8e139a3de715b92023-10-02T23:01:26ZengIEEEIEEE Access2169-35362023-01-011110323210324110.1109/ACCESS.2023.331636510254202Hierarchical Autoencoder Frequency Features for Stress DetectionRadhika Kuttala0https://orcid.org/0000-0001-6133-0749Ramanathan Subramanian1https://orcid.org/0000-0001-9441-7074Venkata Ramana Murthy Oruganti2https://orcid.org/0000-0002-9616-7942Department of Electrical and Electronics Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Coimbatore, IndiaUniversity of Canberra, Bruce, ACT, AustraliaDepartment of Electrical and Electronics Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Coimbatore, IndiaStress has a significant negative impact on people, which has made it a primary social concern. Early stress detection is essential for effective stress management. This study proposes a Deep Learning (DL) method for effective stress detection using multimodal physiological signals - Electrocardiogram (ECG) and Electrodermal activity (EDA). The extensive latent feature representation of DL models has yet to be fully explored. Hence, this paper proposes a hierarchical AutoEncoder (AE) feature fusion on the frequency domain. The latent representations from different layers of the autoencoder are combined and given as input to the classifier - Convolutional Recurrent Neural Network with Squeeze and Excitation (CRNN-SE) model. A two-set performance comparison is performed (romannum 1) performance on frequency band features, and raw data are compared. (romannum 2) autoencoders trained on three cost functions - Mean Squared Error (MSE), Kullback-Leibler (KL) divergence, and Cosine similarity performance are compared on frequency band features and raw data. To verify the generalizability of our approach, we tested it on four benchmark datasets- WAUC, CLAS, MAUS and ASCERTAIN. Results show that frequency band features showed better results than raw data by 4-8%, respectively. MSE loss produced better results than other losses for both frequency band features and raw data by 3-7%, respectively. The proposed approach considerably outperforms existing stress detection models that are subject-independent by 1-2%, respectively.https://ieeexplore.ieee.org/document/10254202/Frequency bandEDAECGstress detectionautoencodershierarchical features |
| spellingShingle | Radhika Kuttala Ramanathan Subramanian Venkata Ramana Murthy Oruganti Hierarchical Autoencoder Frequency Features for Stress Detection IEEE Access Frequency band EDA ECG stress detection autoencoders hierarchical features |
| title | Hierarchical Autoencoder Frequency Features for Stress Detection |
| title_full | Hierarchical Autoencoder Frequency Features for Stress Detection |
| title_fullStr | Hierarchical Autoencoder Frequency Features for Stress Detection |
| title_full_unstemmed | Hierarchical Autoencoder Frequency Features for Stress Detection |
| title_short | Hierarchical Autoencoder Frequency Features for Stress Detection |
| title_sort | hierarchical autoencoder frequency features for stress detection |
| topic | Frequency band EDA ECG stress detection autoencoders hierarchical features |
| url | https://ieeexplore.ieee.org/document/10254202/ |
| work_keys_str_mv | AT radhikakuttala hierarchicalautoencoderfrequencyfeaturesforstressdetection AT ramanathansubramanian hierarchicalautoencoderfrequencyfeaturesforstressdetection AT venkataramanamurthyoruganti hierarchicalautoencoderfrequencyfeaturesforstressdetection |