- An experiment with 25 participants, using an Empatica E4 wristband to record their physiological signals and determine their cognitive stress levels.
- The experimental design can be found on GitHub
- e4_raw_data\preprocessing_e4.ipynb
- preprocessing raw E4 data
- pre-processed and segmented using a sliding window of the length of 30 seconds without overlap
- EDA, BVP, TEMP, HR and 3-axis ACC data from Empatica E4 is used for analysis
- Kfold (classifiers\kfold)
- Leave-One-Subject-Out (LOSO) (classifiers\loso)
- fine-tuning on LOSO (classifiers\loso)
- FCN
- ResNet
- Transformers
- LSTM
- displays dynamic stress fluctuations by utilizing the insights gained from prediction outcomes upon user-specific data
- The application includes a stress meter, which enables users to visually understand their stress levels
-
Maciej Dzieżyc, Martin Gjoreski, Przemysław Kazienko, Stanisław Saganowski, and Matjaž Gams. Can we ditch feature engineering? end-to-end deep learning for affect recognition from physiological sensor data. Sensors, 20(22):6535, 2020
-
Zhiguang Wang, Weizhong Yan, and Tim Oates. Time series classification from scratch with deep neural networks: A strong baseline. In 2017 International joint conference on neural networks (IJCNN), pages 1578–1585. IEEE, 2017
-
Theodoros Ntakouris. https://keras.io/examples/timeseries/, 2021
-
Jürgen Schmidhuber, Sepp Hochreiter, et al. Long short-term memory. Neural Comput, 9(8):1735–1780, 1997
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent