Research map / Probabilistic and scientific modeling / Causal discovery
TRACE: Trajectory Recovery for Continuous Mechanism Evolution in Causal Representation Learning
Shicheng Fan, Kun Zhang, Lu Cheng
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 |
|---|---|---|---|
| constrain mixing coefficients to the simplex instead of unconstrained combinations (ablation) | TRACE · Synthetic OOD transitions | Weight Corr: 0.971 -> 0.956 | Hurt |
| increase observation noise from 0.01 to 0.50 (ablation) | TRACE · Synthetic noise sweep | MCC: 0.996 -> 0.703 | Hurt |
| increase observation noise from 0.01 to 0.50 (ablation) | TRACE · Synthetic noise sweep | Corr: 0.998 -> 0.854 | Hurt |
| increase perturbation magnitude from 0.1 to 0.7 (ablation) | TRACE · Synthetic perturbation sweep | MCC: 0.934 -> 0.990 | Improved |
| increase perturbation magnitude from 0.1 to 0.7 (ablation) | TRACE · Synthetic perturbation sweep | Corr: 0.972 -> 0.998 | Improved |
| add scale calibration (ablation) | TRACE · Synthetic noise sweep (sigma=0.10) | MAE: 0.254 -> 0.030 | Improved |
| increase number of active mechanisms from 2 to 7 (simple) (ablation) | TRACE · Synthetic K_total=10 | alpha recovery corr: .993 -> .776 | Hurt |
| increase number of active mechanisms from 2 to 7 (complex) (ablation) | TRACE · Synthetic K_total=10 | alpha recovery corr: .979 -> .459 | Hurt |
| increase number of active mechanisms from 2 to 7 (simple) (ablation) | TRACE · Synthetic K_total=10 | W recovery corr: 1.000 -> .995 | No clear effect |
| reduce training data from 100% to 10% (ablation) | TRACE · Synthetic sample efficiency | MCC: 0.963 -> 0.752 ± 0.163 | Hurt |
| reduce training data from 100% to 10% (ablation) | TRACE · Synthetic sample efficiency (K=3) | Weight Corr: 0.986 -> 0.960 ± 0.013 | No clear effect |
| reduce training data from 100% to 1% (ablation) | TRACE · Synthetic sample efficiency (K=3) | Weight Corr: 0.986 -> 0.688 ± 0.377 | Hurt |
Closest papers by meaning
- Sim-to-Real Causal Transfer: A Metric Learning Approach to Causally-Aware Interaction Representations (CVPR 2025 · similarity 0.88)
- Act to See, See to Act: Diffusion-Driven Perception-Action Interplay for Adaptive Policies (NeurIPS 2025 · similarity 0.86)
- Fine-Grained Causal Dynamics Learning with Quantization for Improving Robustness in Reinforcement Learning (ICML 2024 · similarity 0.86)
- Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning (ICLR 2025 · similarity 0.86)
- Learning Time-Aware Causal Representation for Model Generalization in Evolving Domains (ICML 2025 · similarity 0.86)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.86)
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.
