Research map / Probabilistic and scientific modeling / Causal discovery
CUTS: Neural Causal Discovery from Irregular Time-Series Data
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Jointly alternate causal graph fitting and learned imputation instead of using a one-shot imputation baseline (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.9206 $\pm$ 0.0216 -> 0.9705 $\pm$ 0.0118 | Helped |
| Add CPG-based parent selection to the imputation stage (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.9535 $\pm$ 0.0167 -> 0.9705 $\pm$ 0.0118 | Helped |
| Add the final fine-tuning stage (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.9665 $\pm$ 0.0096 -> 0.9705 $\pm$ 0.0118 | Helped |
| Replace learned iterative imputation with ZOH imputation (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.8731 $\pm$ 0.0312 -> 0.9705 $\pm$ 0.0118 | Hurt |
| Replace learned iterative imputation with GP imputation (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.8575 $\pm$ 0.0536 -> 0.9705 $\pm$ 0.0118 | Hurt |
| Replace learned iterative imputation with GRIN imputation (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.8983 $\pm$ 0.0299 -> 0.9705 $\pm$ 0.0118 | Hurt |
| Replace ZOH with linear interpolation for initial data filling (ablation) | CUTS · VAR, random missing p=0.6 | AUROC: 0.9766 $\pm$ 0.0074 -> 0.9636 $\pm$ 0.0145 | Hurt |
| Increase the learning budget from 25% to 100% of the original steps (ablation) | CUTS · VAR, random missing p=0.6 | AUROC: 0.9492 $\pm$ 0.0119 -> 0.9766 $\pm$ 0.0074 | Helped |
| Increase the learning budget from 25% to 100% of the original steps (ablation) | CUTS · Lorenz-96, random missing p=0.6 | AUROC: 0.9052 $\pm$ 0.0208 -> 0.9705 $\pm$ 0.0118 | Helped |
| Use CUTS instead of PCMCI with GRIN imputation | Causal discovery on VAR · VAR, random missing p=0.3 | AUROC: 0.9983 $\pm$ 0.0028 -> 0.9971 $\pm$ 0.0026 | Hurt |
| Use CUTS instead of eSRU with GP imputation | Causal discovery on Lorenz-96 · Lorenz-96, random missing p=0.6 | AUROC: 0.9597 $\pm$ 0.0169 -> 0.9705 $\pm$ 0.0118 | Helped |
| Use CUTS instead of NGC with ZOH imputation | Causal discovery on NetSim · NetSim, random missing p=0.2 | AUROC: 0.7668 $\pm$ 0.0403 -> 0.7699 $\pm$ 0.0550 | No clear effect |
Closest papers by meaning
- CUTS+: High-Dimensional Causal Discovery from Irregular Time-Series (AAAI 2024 · similarity 0.96)
- Towards the Disappearing Truth: Fine-Grained Joint Causal Influences Learning with Hidden Variable-Driven Causal Hypergraphs in Time Series (AAAI 2024 · similarity 0.90)
- CausalTime: Realistically Generated Time-series for Benchmarking of Causal Discovery (ICLR 2024 · similarity 0.89)
- TNPAR: Topological Neural Poisson Auto-Regressive Model for Learning Granger Causal Structure from Event Sequences (AAAI 2024 · similarity 0.89)
- GCAD: Anomaly Detection in Multivariate Time Series from the Perspective of Granger Causality (AAAI 2025 · similarity 0.89)
- Learning Causal Relations from Subsampled Time Series with Two Time-Slices (ICML 2024 · similarity 0.88)
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.