Deniz Eroglu: Data-Driven Reconstruction of Brain Network Dynamics
Автор: Machine Learning and Dynamical Systems Seminar
Загружено: 2026-02-23
Просмотров: 37
Описание:
Title: Data-Driven Reconstruction of Brain Network Dynamics
Speaker: Deniz Eroglu
Abstract:
We present a data-driven framework for reconstructing the governing equations and coupling topology of weakly coupled nonlinear systems from limited and noisy time series. Our approach exploits stochastic fluctuations—typically dismissed as noise—as informative signals to infer effective interaction terms and local dynamics via sparse model recovery. Applied to synthetic neuronal networks and experimental recordings from the mouse neocortex, the method accurately recovers both the functional connectivity and dynamical behavior, even with short and partially observed data. This enables forecasting of critical transitions beyond the training regime, offering a tractable route for analyzing high-dimensional biological systems. We conclude with perspectives on incorporating black-box models and normal form constraints to generalize the method to broader classes of complex networks.
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
