Saved in:
Bibliographic Details
Main Authors: Ding, Ziyi, Lai, Xianxin, Chen, Weiyu, Zhang, Xiao-Ping, Chen, Jiayu
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.07712
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911578097451008
author Ding, Ziyi
Lai, Xianxin
Chen, Weiyu
Zhang, Xiao-Ping
Chen, Jiayu
author_facet Ding, Ziyi
Lai, Xianxin
Chen, Weiyu
Zhang, Xiao-Ping
Chen, Jiayu
contents In this work, CausalVAE is introduced as a plug-in structural module for latent world models and is attached to diverse encoder-transition backbones. Across the reported benchmarks, competitive factual prediction is preserved and intervention-aware counterfactual retrieval is improved after the plug-in is added, suggesting stronger robustness under distribution shift and interventions. The largest gains are observed on the Physics benchmark: when averaged over 8 paired baselines, CF-H@1 is improved by +102.5%. In a representative GNN-NLL setting on Physics, CF-H@1 is increased from 11.0 to 41.0 (+272.7%). Through causal analysis, learned structural dependencies are shown to recover meaningful first-order physical interaction trends, supporting the interpretability of the learned latent causal structure.
format Preprint
id arxiv_https___arxiv_org_abs_2604_07712
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CausalVAE as a Plug-in for World Models: Towards Reliable Counterfactual Dynamics
Ding, Ziyi
Lai, Xianxin
Chen, Weiyu
Zhang, Xiao-Ping
Chen, Jiayu
Machine Learning
In this work, CausalVAE is introduced as a plug-in structural module for latent world models and is attached to diverse encoder-transition backbones. Across the reported benchmarks, competitive factual prediction is preserved and intervention-aware counterfactual retrieval is improved after the plug-in is added, suggesting stronger robustness under distribution shift and interventions. The largest gains are observed on the Physics benchmark: when averaged over 8 paired baselines, CF-H@1 is improved by +102.5%. In a representative GNN-NLL setting on Physics, CF-H@1 is increased from 11.0 to 41.0 (+272.7%). Through causal analysis, learned structural dependencies are shown to recover meaningful first-order physical interaction trends, supporting the interpretability of the learned latent causal structure.
title CausalVAE as a Plug-in for World Models: Towards Reliable Counterfactual Dynamics
topic Machine Learning
url https://arxiv.org/abs/2604.07712