license: mit
language:
- en
Pre-trained Models for EDA, PPG, and EDA+PPG
This repository provides pre-trained CLSP models for the EEVR dataset and the FEEL benchmarking framework.
If you use these models in your research, please cite the following papers:
EEVR (NeurIPS 2024) https://proceedings.neurips.cc/paper_files/paper/2024/hash/1cba8502063fab9df252a63968691768-Abstract-Datasets_and_Benchmarks_Track.html
FEEL (NeurIPS 2025) Quantifying Heterogeneity in Physiological Signals for Generalizable Emotion Recognition https://neurips.cc/virtual/2025/loc/san-diego/poster/121715 Model Details
The released models support the following physiological modalities:
Electrodermal Activity (EDA), Photoplethysmography (PPG), Multimodal (EDA + PPG)
They are trained for:
2-class classification (arousal, valence), 4-class classification (Russell's (1980) Circumplex Models)
Usage:
You can load and experiment with these models using the accompanying code and resources available in the original repositories (see below). Note: The 2-class results reported in the original EEVR paper were obtained using Leave-One-Subject-Out (LOSO) cross-validation. For the FEEL work, the 2-class models were retrained on the full EEVR dataset.
References:
EEVR Repository: https://github.com/alchemy18/EEVR
FEEL Repository: https://github.com/alchemy18/FEEL