# Artifact Annotations Partially Substitute for Per-User Calibration: SAFE-EDA and a Normalization-Controlled Evaluation of Wrist-EDA Affect Recognition

arXiv: https://arxiv.org/abs/2610.01692 · HTML: https://mlsanity.com/paper/2610.01692/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 77 references that are in the map.
- First posted on arXiv 1 Oct 2026.

## Builds on (most cited first, 11 of 11)

- [Improving Reproducibility in Machine Learning Research (A Report from the NeurIPS 2019 Reproducibility Program)](https://mlsanity.com/paper/2003.12206/) (2020)
- [A Comparison of Personalized and Generalized Approaches to Emotion Recognition Using Consumer Wearable Devices: Machine Learning Study](https://mlsanity.com/paper/2308.14245/) (2023)
- [Ensemble Machine Learning Model Trained on a New Synthesized Dataset Generalizes Well for Stress Prediction Using Wearable Devices](https://mlsanity.com/paper/2209.15146/) (2022)
- [Unsupervised Motion Artifact Detection in Wrist-Measured Electrodermal Activity Data](https://mlsanity.com/paper/1707.08287/) (2017)
- [Stress Assessment with Convolutional Neural Network Using PPG Signals](https://mlsanity.com/paper/2410.12273/) (2024)
- [SELF-CARE: Selective Fusion with Context-Aware Low-Power Edge Computing for Stress Detection](https://mlsanity.com/paper/2205.03974/) (2022)
- [Stress Detection using Context-Aware Sensor Fusion from Wearable Devices](https://mlsanity.com/paper/2303.08215/) (2023)
- [Real-Time Stress Detection via Photoplethysmogram Signals: Implementation of a Combined Continuous Wavelet Transform and Convolutional Neural Network on Resource-Constrained Microcontrollers](https://mlsanity.com/paper/2410.19776/) (2024)
- [Cross-Modality Investigation on WESAD Stress Classification](https://mlsanity.com/paper/2502.18733/) (2025)
- [A Preliminary Study on Automatic Motion Artifacts Detection in Electrodermal Activity Data Using Machine Learning](https://mlsanity.com/paper/2107.07650/) (2021)
- [Stress Classification from ECG Signals Using Vision Transformer](https://mlsanity.com/paper/2603.26721/) (2026)

## Abstract (arXiv)

Wrist electrodermal activity (EDA) differs in amplitude from one person to the next, so affect-recognition models normalize their input before classification. Studies that test such models on held-out subjects seldom report where the normalization statistics come from, yet statistics computed from the held-out subject's own recording give the model information that a device does not have when it is first worn. We asked how this choice alters the measured benefit of pretraining. A compact convolutional network, SAFE-EDA, was pretrained on expert artifact annotations from 43 subjects and compared with the same network trained from scratch on the Wearable Stress and Affect Detection (WESAD) dataset (15 subjects, leave-one-subject-out), with two normalization sources crossed with four window hops. When the statistics came only from training subjects, pretraining raised macro-F1 by 0.078 to 0.227; when they came from the held-out user's full recording, the gain fell to between 0.020 and 0.050 and was no longer significant. Artifact supervision was far more useful than self-supervised pretraining on the same recordings (0.078 versus 0.008). Across 13 configurations in two datasets, the pretrained network was better in 12, but on the second dataset (26 subjects) per-user normalization increased the gain instead of reducing it, so the interaction depends on the data. Only five of 50 published WESAD studies state which data were used for normalization. Reporting this choice is necessary to separate first-use performance from performance after calibration.