LearningContinuousSystemDynamicsfrom Irregularly-SampledPartialObservations

Our model employs anovel encoder parameterized by a graph neural network that can infer initial states in an unsupervised way from irregularly-sampled partial observations of structural objects and utilizes neural ODEtoinferarbitrarily complexcontinuous-time latentdynamics. Experiments onmotion capture, spring system, and charged particle datasets demonstrate the effectivenessofourapproach.