Timezone: »
Data-driven approaches to modeling physical systems fail to generalize to unseen systems that share the same general dynamics with the learning domain, but correspond to different physical contexts. We propose a new framework for this key problem, context-informed dynamics adaptation (CoDA), which takes into account the distributional shift across systems for fast and efficient adaptation to new dynamics. CoDA leverages multiple environments, each associated to a different dynamic, and learns to condition the dynamics model on contextual parameters, specific to each environment. The conditioning is performed via a hypernetwork, learned jointly with a context vector from observed data. The proposed formulation constrains the search hypothesis space to foster fast adaptation and better generalization across environments. It extends the expressivity of existing methods. We theoretically motivate our approach and show state-of-the-art generalization results on a set of nonlinear dynamics, representative of a variety of application domains. We also show, on these systems, that new system parameters can be inferred from context vectors with minimal supervision.
Author Information
Matthieu Kirchmeyer (Sorbonne Université & Criteo AI Lab)
Yuan Yin (Sorbonne Université, ISIR, MLIA team)
Jérémie DONA (Sorbonne Université)
Nicolas Baskiotis (LIP6)
alain rakotomamonjy (Criteo)
Patrick Gallinari (Sorbonne Universite, Criteo AI Lab)
Related Events (a corresponding poster, oral, or spotlight)
-
2022 Poster: Generalizing to New Physical Systems via Context-Informed Dynamics Model »
Tue. Jul 19th through Wed the 20th Room Hall E
More from the Same Authors
-
2022 : A Bias-Variance Analysis of Weight Averaging for OOD Generalization »
Alexandre Ramé · Matthieu Kirchmeyer · Thibaud J Rahier · Alain Rakotomamonjy · Patrick Gallinari · Matthieu Cord -
2022 Poster: A Neural Tangent Kernel Perspective of GANs »
Jean-Yves Franceschi · Emmanuel de Bézenac · Ibrahim Ayed · Mickael Chen · Sylvain Lamprier · Patrick Gallinari -
2022 Spotlight: A Neural Tangent Kernel Perspective of GANs »
Jean-Yves Franceschi · Emmanuel de Bézenac · Ibrahim Ayed · Mickael Chen · Sylvain Lamprier · Patrick Gallinari -
2021 Poster: Differentially Private Sliced Wasserstein Distance »
alain rakotomamonjy · Ralaivola Liva -
2021 Oral: Differentially Private Sliced Wasserstein Distance »
alain rakotomamonjy · Ralaivola Liva -
2020 Poster: Partial Trace Regression and Low-Rank Kraus Decomposition »
Hachem Kadri · Stephane Ayache · Riikka Huusari · alain rakotomamonjy · Ralaivola Liva -
2019 Poster: Screening rules for Lasso with non-convex Sparse Regularizers »
alain rakotomamonjy · Gilles Gasso · Joseph Salmon -
2019 Oral: Screening rules for Lasso with non-convex Sparse Regularizers »
alain rakotomamonjy · Gilles Gasso · Joseph Salmon