Generate your neural signals from mine: individual-to-individual EEG converters

• DOI: 10.48550/arxiv.2304.10736
• Publication Date: 2023-01-01
• Authors: Zitong Lu; Julie D. Golomb
• Keywords: Generative model

Most models in cognitive and computational neuroscience trained on one subject do not generalize to other subjects due individual differences. An ideal individual-to-individual neural converter is expected generate real signals of from those another one, which can overcome the problem differences for models. In this study, we propose a novel EEG converter, called EEG2EEG, inspired by generative computer vision. We applied THINGS EEG2 dataset train test 72 independent EEG2EEG corresponding pairs across 9 subjects. Our results demonstrate that able effectively learn mapping representations achieve high conversion performance. Additionally, generated contain clearer visual information than be obtained data. This method establishes state-of-the-art framework signals, realize flexible high-performance provide insight both engineering neuroscience.