Disentangled generative models for robust dynamical system prediction

Stathi Fotiadis, Shunlong Hu, Mario Lino Valencia, Chris D Cantwell, Anil Anthony Bharath

Keywords: disentanglement, dynamical systems, prediction, generative models, robustness, out-of-distribution, distribution shift, causal inference

Abstract: Deep neural networks have become increasingly of interest in dynamical system prediction, but out-of-distribution generalization and long-term stability still remains challenging. In this work, we treat the domain parameters of dynamical systems as factors of variation of the data generating process. By leveraging ideas from supervised disentanglement and causal factorization, we aim to separate the domain parameters from the dynamics in the latent space of generative models. In our experiments we model dynamics both in phase space and in video sequences and conduct rigorous OOD evaluations. Results indicate that disentangled models adapt better to domain parameters spaces that were not present in the training data while, at the same time, provide better long-term predictions in video sequences.

One-sentence Summary: Disentangling ODE parameters from dynamics leads to better long-term and out-of-distribution predictions