Schedule of Events | Symposia

Removing neural signal artifacts with autoencoder-targeted adversarial transformers (AT-AT)

Poster Session F - Tuesday, April 1, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

Benjamin Choi¹ (benchoi@college.harvard.edu); ¹Harvard University

Electromyogenic (EMG) noise is a major contamination source in EEG data that can impede accurate analysis of brain-specific neural activity. Recent literature on EMG artifact removal has moved beyond traditional linear algorithms in favor of machine learning-based systems. However, existing deep learning-based filtration methods often have large compute footprints and prohibitively long training times. In this study, we present a new machine learning-based system for filtering EMG interference from EEG data using an autoencoder-targeted adversarial transformer (AT-AT). By leveraging the lightweight expressivity of an autoencoder to determine optimal time-series transformer application sites, our AT-AT architecture achieves a >90% model size reduction compared to published artifact removal models. The addition of adversarial training ensures that filtered signals adhere to the fundamental characteristics of EEG data. We trained AT-AT using published neural data from 67 subjects and found that the system was able to achieve comparable test performance to larger models; AT-AT posted a mean reconstructive correlation coefficient above 0.95 at an initial signal-to-noise ratio (SNR) of 2 dB and 0.70 at -7 dB SNR. We then further tested our system by deploying AT-AT in conjunction with a downstream classifier on a notable inferential use case: distinguishing imagined digits and non-digits from EEG data. We found that AT-AT filtration helped reduce classification error by over 40% and increased latent space separability of neural signal classes. Our results demonstrate the potential of AT-AT for streamlined artifact removal; further research generalizing these results to broader sample sizes and additional use cases will be crucial.

Topic Area: METHODS: Electrophysiology