%0 Journal Article %1 saemundsson2019variational %A Saemundsson, Steindor %A Terenin, Alexander %A Hofmann, Katja %A Deisenroth, Marc Peter %D 2019 %K bayesian readings stochastic variational %T Variational Integrator Networks for Physically Meaningful Embeddings %U http://arxiv.org/abs/1910.09349 %X Learning workable representations of dynamical systems is becoming an increasingly important problem in a number of application areas. By leveraging recent work connecting deep neural networks to systems of differential equations, we propose variational integrator networks, a class of neural network architectures designed to ensure faithful representations of the dynamics under study. This class of network architectures facilitates accurate long-term prediction, interpretability, and data-efficient learning, while still remaining highly flexible and capable of modeling complex behavior. We demonstrate that they can accurately learn dynamical systems from both noisy observations in phase space and from image pixels within which the unknown dynamics are embedded.

@article{saemundsson2019variational, abstract = {Learning workable representations of dynamical systems is becoming an increasingly important problem in a number of application areas. By leveraging recent work connecting deep neural networks to systems of differential equations, we propose variational integrator networks, a class of neural network architectures designed to ensure faithful representations of the dynamics under study. This class of network architectures facilitates accurate long-term prediction, interpretability, and data-efficient learning, while still remaining highly flexible and capable of modeling complex behavior. We demonstrate that they can accurately learn dynamical systems from both noisy observations in phase space and from image pixels within which the unknown dynamics are embedded.}, added-at = {2019-10-22T13:56:22.000+0200}, author = {Saemundsson, Steindor and Terenin, Alexander and Hofmann, Katja and Deisenroth, Marc Peter}, biburl = {https://www.bibsonomy.org/bibtex/25def13e75821673a2b0439cb0f5a53ef/kirk86}, description = {[1910.09349] Variational Integrator Networks for Physically Meaningful Embeddings}, interhash = {419f6e40d582b6a7a30a12fa937e777d}, intrahash = {5def13e75821673a2b0439cb0f5a53ef}, keywords = {bayesian readings stochastic variational}, note = {cite arxiv:1910.09349}, timestamp = {2019-10-22T14:00:42.000+0200}, title = {Variational Integrator Networks for Physically Meaningful Embeddings}, url = {http://arxiv.org/abs/1910.09349}, year = 2019 }