Wearable Abnormal Emotion Detection

This repository contains the code for the paper "Seamless Monitoring of Stress Levels Leveraging a Universal Model for Time Sequences".

Results

Comparison of Different Methods

	DREAMER	HCI	WESAD (ECG)	WESAD (BVP)	AVG F1
LSTM-NDT	0.313	0.375	0.785	0.772	0.561 ± .218
OmniAnomaly	0.599	0.547	0.767	0.650	0.641 ± .081
CAE-M	0.658	0.530	0.801	0.590	0.645 ± .101
HypAD	0.650	0.643	0.815	0.567	0.669 ± .090
MTAD-GAT	0.536	0.742	0.857	0.719	0.714 ± .115
TranAD	0.617	0.623	0.837	0.804	0.720 ± .101
TadGAN	0.590	0.691	0.864	0.743	0.722 ± .099
GDN	0.713	0.580	0.858	0.802	0.738 ± .105
DAGMM	0.743	0.647	0.831	0.773	0.749 ± .067
USAD	0.730	0.660	0.830	0.797	0.754 ± .065
MAD-GAN	0.706	0.743	0.839	0.787	0.769 ± .050
MSCRED	0.675	0.824	0.876	0.775	0.788 ± .074
UniTS	0.869	0.878	0.834	0.856	0.859 ± .019

P-values from Dunn Post-hoc Test

	UniTS
MSCRED	$2.813 \times 10^{-2}$
MAD-GAN	$4.795 \times 10^{-3}$
USAD	$6.839 \times 10^{-3}$
DAGMM	$1.136 \times 10^{-3}$
GDN	$5.146 \times 10^{-3}$

Setup

The following command installs the required dependencies:

# Install the dependencies
pip install -r requirements.txt

Datasets and Preprocessing

The following datasets are used in this project:

• WESAD
• DREAMER
• MAHNOB-HCI

The datasets are preprocessed using the following steps:

# Download and extract the datasets
./data/download-datasets.sh

# Preprocess the datasets
python preprocess.py

Training

Ensure to be logged and to set your WANDB_USER in the file run.py to log the results to Weights & Biases. The following command launches the training of the models:

# Train the models
python run.py

Evaluation

The evaluation results are processed from logged runs on Weights & Biases. To retrieve the evaluation results, run the following command:

# Evaluate the models
python results.py

Acknowledgements

This project use the code from the repositiories of UniTS, TranAD and HypAD for the implementation of the models.