From Signals to Graphs: Recent Advances in EEG Representation Learning (Nov 18)

Speaker: Yuyang Dai

Date: Nov 18, 11:45 AM – 12:45 PM

Abstract:

Electroencephalography (EEG) has long served as a powerful tool for understanding human brain activity, yet traditional signal-based analysis struggles to capture the complex, dynamic, and non-Euclidean nature of neural interactions. Recent progress in deep learning-particularly in graph neural networks (GNNs) and foundation model paradigms-has transformed how EEG data are represented and interpreted. This seminar provides a comprehensive overview of emerging approaches that move beyond sequential processing toward graph-based and self-supervised representations of brain dynamics. We first review how spatial, functional, and learnable graph construction methods unify EEG signals into interpretable brain connectivity structures. Then, we highlight the evolution from static GNNs to dynamic and geometric models capable of tracking temporal brain state transitions. Further, we discuss how transfer learning and pre-training frameworks, such as Graph Contrastive Autoencoders and EEG foundation models, enable cross-subject and cross-dataset generalization. Finally, we explore how explainable graph learning bridges data-driven analysis and clinical interpretability, supporting applications in stress decoding, epilepsy detection, motor imagery, and neurological disorder classification. By tracing the methodological shift from raw signals to structured graphs and foundation models, this seminar aims to provide an integrative understanding of how modern representation learning redefines EEG analysis and moves us closer to generalizable, explainable, and clinically reliable brain decoding. To help you better understand the recent literature, you can find summaries of high-citation and top-conference EEG papers: HERE

Speaker Profile

I am currently a Research Assistant at the Singapore University of Technology and Design under the Computer Science and Design pillar, working with Prof. Wenxuan Zhang. My current research focuses on graph-based neural modeling, and large foundation models for complex real-world decision-making. I am particularly interested in how graph representation learning and self-supervised techniques can bridge low-level neural dynamics with high-level cognitive or behavioral understanding. In 2026 Fall, I will join the Ph.D. program in Computer Science at Florida State University, where I plan to continue exploring scalable, interpretable, and generalizable graph learning frameworks for safety and neural system areas.

Location: LOV 307 (In Person Only)