Saved in:
Bibliographic Details
Main Authors: Cheng, Sheng, Kong, Deqian, Xie, Jianwen, Lee, Kookjin, Wu, Ying Nian, Yang, Yezhou
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.03845
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929698574958592
author Cheng, Sheng
Kong, Deqian
Xie, Jianwen
Lee, Kookjin
Wu, Ying Nian
Yang, Yezhou
author_facet Cheng, Sheng
Kong, Deqian
Xie, Jianwen
Lee, Kookjin
Wu, Ying Nian
Yang, Yezhou
contents This paper introduces novel deep dynamical models designed to represent continuous-time sequences. Our approach employs a neural emission model to generate each data point in the time series through a non-linear transformation of a latent state vector. The evolution of these latent states is implicitly defined by a neural ordinary differential equation (ODE), with the initial state drawn from an informative prior distribution parameterized by an Energy-based model (EBM). This framework is extended to disentangle dynamic states from underlying static factors of variation, represented as time-invariant variables in the latent space. We train the model using maximum likelihood estimation with Markov chain Monte Carlo (MCMC) in an end-to-end manner. Experimental results on oscillating systems, videos and real-world state sequences (MuJoCo) demonstrate that our model with the learnable energy-based prior outperforms existing counterparts, and can generalize to new dynamic parameterization, enabling long-horizon predictions.
format Preprint
id arxiv_https___arxiv_org_abs_2409_03845
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Latent Space Energy-based Neural ODEs
Cheng, Sheng
Kong, Deqian
Xie, Jianwen
Lee, Kookjin
Wu, Ying Nian
Yang, Yezhou
Machine Learning
This paper introduces novel deep dynamical models designed to represent continuous-time sequences. Our approach employs a neural emission model to generate each data point in the time series through a non-linear transformation of a latent state vector. The evolution of these latent states is implicitly defined by a neural ordinary differential equation (ODE), with the initial state drawn from an informative prior distribution parameterized by an Energy-based model (EBM). This framework is extended to disentangle dynamic states from underlying static factors of variation, represented as time-invariant variables in the latent space. We train the model using maximum likelihood estimation with Markov chain Monte Carlo (MCMC) in an end-to-end manner. Experimental results on oscillating systems, videos and real-world state sequences (MuJoCo) demonstrate that our model with the learnable energy-based prior outperforms existing counterparts, and can generalize to new dynamic parameterization, enabling long-horizon predictions.
title Latent Space Energy-based Neural ODEs
topic Machine Learning
url https://arxiv.org/abs/2409.03845