Research map / Probabilistic and scientific modeling / Causal discovery
One Intervention per Component is Enough: Towards Identifiability in Linear Stochastic Dynamics from Steady State
Saber Salehkaleybar
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 mean-equation residuals to the covariance-only loss (ablation) | Regularized least-squares OU estimator · Synthetic OU, n=10 | relative error of Lambda, D, b | Improved |
| add mean-equation residuals to covariance-only loss (ablation) | OU estimator on Perturb-seq · Perturb-seq Co-Culture | PDS: 0.43 -> 0.67 | Improved |
| add mean-equation residuals to covariance-only loss (ablation) | OU estimator on Perturb-seq · Perturb-seq Co-Culture | DES: 0.51 -> 0.43 | Hurt |
| add mean-equation residuals to covariance-only loss (ablation) | OU estimator on Perturb-seq · Perturb-seq Control | DES: 0.46 -> 0.33 | Hurt |
| add mean-equation residuals to covariance-only loss (ablation) | OU estimator on Perturb-seq · Perturb-seq IFN-gamma | PDS: 0.43 -> 0.70 | Improved |
| add mean-equation residuals to covariance-only loss (ablation) | OU estimator on Perturb-seq · Perturb-seq IFN-gamma | DES: 0.42 -> 0.47 | Improved |
| increase the number of interventions from 0 to 10 | Regularized least-squares OU estimator · Non-diagonal diffusion synthetic | relative error of Lambda: 0.2064 -> 0.0646 | Improved |
| increase the number of interventions from 0 to 10 | Regularized least-squares OU estimator · Non-diagonal diffusion synthetic | relative error of D: 0.1811 -> 0.0719 | Improved |
| increase the number of interventions from 0 to 10 | Regularized least-squares OU estimator · Non-diagonal diffusion synthetic | relative error of b: 0.1407 -> 0.0661 | Improved |
| increase the number of interventions on diagonal-diffusion D estimate | Regularized least-squares OU estimator · Synthetic OU, n=10 | relative error of D | Mixed |
| increase sample size from 500 to 10000 for hard interventions | SCC recovery from mean shifts · Synthetic SCC recovery | Adjusted Rand Index: 0.660 -> 0.881 | Improved |
| increase sample size from 500 to 10000 for soft interventions | SCC recovery from mean shifts · Synthetic SCC recovery | Adjusted Rand Index: 0.464 -> 0.837 | Improved |
Closest papers by meaning
- On the Identifiability of Switching Dynamical Systems (ICML 2024 · similarity 0.87)
- Disentangling Continuous-Time Latent Dynamics: Identifiability of Latent SDEs via Diffusion Shifts (NeurIPS 2026 · similarity 0.87)
- Identifiability Analysis of Linear ODE Systems with Hidden Confounders (NeurIPS 2024 · similarity 0.86)
- SING: SDE Inference via Natural Gradients (NeurIPS 2025 · similarity 0.85)
- Learning non-equilibrium diffusions with Schrödinger bridges: from exactly solvable to simulation-free (NeurIPS 2025 · similarity 0.85)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.85)
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.
