Keywords: Visual Evoked Potential, Brain-machine Interface, EEGs, Closed-loop AI, Generative AI
TL;DR: "The EEG Engine" uses a DCGAN and EEG decoder to autonomously generate visual stimuli, improving EEG biomarker reliability for brain-machine interfaces with adaptive, real-time neural feedback, validated by SSVEP tests.
Abstract: The effectiveness of Visual Brain-Machine Interfaces (BMIs) is significantly dependent on the accurate detection and interpretation of electroencephalography (EEG) biomarkers, which frequently exhibit variability due to physiological changes and environmental disturbances over time. Traditional EEG signal enhancement strategies largely concentrate on signal processing techniques such as feature extraction and filtering; however, these approaches often do not adequately address the inherent sources of variability that affect biomarker stability over time. To surmount these challenges, we have developed the Visual Evoked Potential Booster (VEP Booster), a novel closed-loop artificial intelligence framework designed to produce reliable and stable EEG biomarkers under visual stimulation protocols. Our system utilizes a Deep Convolutional Generative Adversarial Network (DCGAN) to refine stimulus images based on real-time feedback from human EEG signals, thereby creating visual stimuli that are specifically tailored to the characteristic preferences of neurons in the primary visual cortex. We evaluated the efficacy of this system through the implementation of steady-state visual evoked potential (SSVEP) protocols in nine human subjects. In our evaluations, both the SSVEP biomarker amplitude and the single-trial SSVEP binary classification experiments, encompassing intra- and inter-temporal analyses, exhibited statistically significant enhancements when employing the VEP Booster. These encouraging outcomes underscore the potential for broad applications in clinical and technological domains.
Primary Area: applications to neuroscience & cognitive science
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 9208
Loading