Evaluating Latent Space Robustness and Uncertainty of EEG-ML Models under Realistic Distribution Shifts
Abstract
The recent availability of large datasets in bio-medicine has inspired the development of representation learning methods for multiple healthcare applications. Despite advances in predictive performance, the clinical utility of such methods is limited when exposed to real-world data. This study develops model diagnostic measures to detect potential pitfalls before deployment without assuming access to external data. Specifically, we focus on modeling realistic data shifts in electrophysiological signals (EEGs) via data transforms and extend the conventional task-based evaluations with analyses of a) the model's latent space and b) predictive uncertainty under these transforms. We conduct experiments on multiple EEG feature encoders and two clinically relevant downstream tasks using publicly available large-scale clinical EEGs. Within this experimental setting, our results suggest that measures of latent space integrity and model uncertainty under the proposed data shifts may help anticipate performance degradation during deployment.
Keywords
Cite
@article{arxiv.2209.11233,
title = {Evaluating Latent Space Robustness and Uncertainty of EEG-ML Models under Realistic Distribution Shifts},
author = {Neeraj Wagh and Jionghao Wei and Samarth Rawal and Brent M. Berry and Yogatheesan Varatharajah},
journal= {arXiv preprint arXiv:2209.11233},
year = {2022}
}
Comments
NeurIPS 2022 camera ready version. Code available at https://github.com/neerajwagh/evaluating-eeg-representations. tl;dr - We develop model diagnostic measures to identify failure modes of EEG-ML models before deployment without access to out-of-distribution data. Keywords - dataset shift, EEG, representation learning, robustness, latent space, uncertainty quantification, distribution shift