Meta-iLaD: Identifiable Latent Dynamics via Meta-Learning of Dynamics Environments
Yubo Ye ⋅ Sweekar Piya ⋅ Xiajun Jiang ⋅ Linwei Wang
Abstract
Learning *latent dynamics* is central to assessing current states and forecasting future trajectories for high-dimensional time series. For locally-stationary latent dynamics of the form $\mathcal{F}(\mathbf{z}_{
Successful Page Load