Seminar: Normalizing Flows and Invertible Dynamic Systems, 29th August, 1:00pm
When: Thursday 29th of August, 1:00pm AEST
Where: This seminar will be partially presented at the J18 lvl 4 Conference Room and partially online via Zoom. RSVP
Speaker: Yurui Zhang
Title: Normalizing Flows and Invertible Dynamic Systems
Abstract:
Normalizing flows is a type of generative models that utilize invertible mappings to transform a simple base distribution into a more complex target distribution, while predominately deal with finite-dimensional static dataset . In robotics, learning and predicting with infinite-dimensional time series data is important for tasks including imitation learning, trajectory generation, temporal forecasting, etc. This talk will briefly introduce the idea of normalizing flow and invertible dynamic systems, and motivated from that, propose a novel invertible recurrent network framework with guaranteed stability and passivity to map from signal space to signal space.
Bio:
Yurui Zhang received his bachelor’s degree in mechanical engineering at The University of Sydney and started his PhD research at ACFR in 2021 working on invertible neural networks. His research interest incudes robust machine learning and nonlinear control.