Research map / Probabilistic and scientific modeling / Causal discovery
CausalVAE as a Plug-in for World Models: Towards Reliable Counterfactual Dynamics
Ziyi Ding, Xianxin Lai, Weiyu Chen, Xiao-Ping Zhang, Jiayu Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add CausalVAE layer | GNN_NLL · Physics 3-body | CF-H@1: 11.0 -> 41.0 | Improved |
| add CausalVAE layer | AE_NLL · Physics 3-body | CF-H@1: 10.50 -> 41.00 | Improved |
| add CausalVAE layer | VAE_Contrastive · Physics 3-body | CF-H@1: 8.50 -> 8.50 | No clear effect |
| add CausalVAE layer | AE_Contrastive · Chemistry | CF-H@1: 52.61 -> 35.87 | Hurt |
| add CausalVAE layer | GNN_NLL · 2D Shapes | CF-H@1: 33.00 -> 54.75 | Improved |
| add CausalVAE layer | Modular_Contrastive · 3D Cubes | CF-H@1: 7.90 -> 16.75 | Improved |
| add CausalVAE layer | Modular_Contrastive · Chemistry | H@1 (1 step): 99.50 -> 16.03 | Hurt |
| add CausalVAE layer | GNN_Contrastive · Physics 3-body | CF-H@1: 11.50 -> 15.00 | Mixed |
| add three-stage training instead of joint training (ablation) | CausalVAE · Physics ablation (GNN_Contrastive backbone) | CF-H@1: 28.00 -> 31.00 | Improved |
| add CausalVAE branch (ablation) | GNN_Contrastive · Physics ablation (GNN_Contrastive backbone) | H@1: 80.00 -> 91.00 | Improved |
| add state alignment loss (ablation) | CausalVAE · Physics ablation (GNN_Contrastive backbone) | H@1: 9.00 -> 91.00 | Improved |
| turn fusion gate on (ablation) | CausalVAE · Physics ablation (GNN_Contrastive backbone) | CF-H@1: 13.67 -> 31.00 | Improved |
Closest papers by meaning
- Learning Structural Causal Models from Ordering: Identifiable Flow Models (AAAI 2025 · similarity 0.88)
- Towards Characterizing Domain Counterfactuals for Invertible Latent Causal Models (ICLR 2024 · similarity 0.88)
- Causal Representation Learning via Counterfactual Intervention (AAAI 2024 · similarity 0.88)
- CausalTab: Pretraining Across Causal Environments for Tabular Causal Discovery (NeurIPS 2026 · similarity 0.87)
- Curious Causality-Seeking Agents Learn Meta Causal World (NeurIPS 2025 · similarity 0.87)
- A New Approach to Backtracking Counterfactual Explanations: A Unified Causal Framework for Efficient Model Interpretability (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
