Geometric methods for inference of nonlinear systems
Date:
Twenty minutes seminar talk at the group meeting of the Gjiorgjieva lab at the School of Life Sciences at the Technical University of Munich.
Presented two frameworks on dynamical inference with geometric considerations:
- One from the before my PhD thesis work on inference of synaptic connectivity from spike trains relying on a geometric approximation of the inter-spike interval generating function of recorded neurons (Casadiego*, Maoutsa*, Timme; PRL; 2018).
- One from my PhD thesis work on inference of sparsely observed stochastic dynamical systems with geometric inductive biases (Maoutsa; MLPS-NeurIPS; 2022, Maoutsa; Physics4ML-ICLR; 2023). (Short intuitve description due to the unfamiliarity of most of the audience with stochastic dynamics and path integrals.)
Invited by Julijana Gjiorgjieva.