Research map / Probabilistic and scientific modeling / Causal discovery
TCD-Arena: Assessing Robustness of Time Series Causal Discovery Methods Against Assumption Violations
Gideon Stein, Niklas Penzel, Tristan Piater, Joachim Denzler
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 |
|---|---|---|---|
| Use VARLiNGAM instead of other CD methods for lagged graph (correctly specified lag) | VARLiNGAM · TCD-Arena, G_LWCG | total robustness | Improved |
| Restrict model max lag below the true lag (ablation) | All CD methods · TCD-Arena, G_LWCG | avg normalized SHD: Varlingam .784 (-.38); GVAR .782 (-.36) | Hurt |
| Overestimate the model max lag (ablation) | All CD methods · TCD-Arena, G_LWCG | avg normalized SHD: Varlingam .429 (-.02); GVAR .467 (-.04) | No clear effect |
| Use non-Gaussian (uniform/Weibull) innovation noise with VARLiNGAM (ablation) | VARLiNGAM · TCD-Arena, innovation noise violations | robustness | No clear effect |
| Average robustness over all hyperparameters instead of the best one, for methods with large hyperparameter sets (ablation) | Dynotears, Nts-Notears · TCD-Arena | avg normalized SHD: Nts-Notears .634 ± .18; Dynotears .507 ± .06 | Hurt |
| Add a linear meta-learner ensemble over base CD methods | CD ensemble · TCD-Arena, G_LWCG and G_LSG | robustness | Improved |
| Use MLP/Transformer meta-learner ensembles | CD ensemble · TCD-Arena | robustness | No clear effect |
| Use ensembles for instantaneous graph | CD ensemble · TCD-Arena, G_INST | robustness | No clear effect |
Closest papers by meaning
- CausalTime: Realistically Generated Time-series for Benchmarking of Causal Discovery (ICLR 2024 · similarity 0.89)
- THE ROBUSTNESS OF DIFFERENTIABLE CAUSAL DISCOVERY IN MISSPECIFIED SCENARIOS (ICLR 2025 · similarity 0.89)
- CausalRivers - Scaling up benchmarking of causal discovery for real-world time-series (ICLR 2025 · similarity 0.89)
- CausalDynamics: A large‐scale benchmark for structural discovery of dynamical causal models (NeurIPS 2025 · similarity 0.88)
- Challenges and Considerations in the Evaluation of Bayesian Causal Discovery (ICML 2024 · similarity 0.87)
- On the Recoverability of Causal Relations from Temporally Aggregated I.I.D. Data (ICML 2024 · 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.
