Research map / Probabilistic and scientific modeling / Variational and generative sampling
Conditional Dependence via Shannon Capacity: Axioms, Estimators and Applications
Weihao Gao, Sreeram Kannan, Sewoong Oh, Pramod Viswanath
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 CMI (Shannon capacity with optimized input weights) instead of UMI (uniform input) as the causal strength measure | conditional dependence estimation on flow-cytometry data · T-cell flow-cytometry trend recovery under subsampling | fraction of subsamples recovering true biological trend: uniformly dominates UMI at all subsample sizes (qualitative) | Improved |
| Replace DREMI-style partition/KDE estimation with UMI kNN importance-sampling estimator | UMI estimation on flow-cytometry data · T-cell flow-cytometry trend recovery under subsampling | fraction of subsamples recovering true biological trend: uniformly dominates DREMI when fewer samples available | Improved |
| Use UMI/CMI estimators on full flow-cytometry dataset | UMI and CMI vs DREMI · Full flow-cytometry dataset | recovery of biological trend | No clear effect |
| Use UMI as direction-of-causality score | causal direction benchmark · Causal direction benchmark datasets | accuracy of predicting correct direction: 45% vs CMI 53%, marginal entropy 45% | No clear effect |
| Use CMI as direction-of-causality score | causal direction benchmark · Causal direction benchmark datasets | accuracy of predicting correct direction: 53% (near chance) | No clear effect |
| Replace partition-based estimator with proposed UMI kNN estimator | UMI estimation on synthetic Beta/Gaussian data · Synthetic data, sigma^2 in {0.09,0.36,1.0} | sample complexity to match ground truth | Improved |
| Replace Blahut-Arimoto on quantized data with proposed CMI kNN optimization estimator | CMI estimation on synthetic data · Synthetic data, sigma^2 in {0.36,1.0,2.25} | sample complexity | Improved |
| Add nearest-neighbor regularization of rho_{k,i} (rho^{d}>Ck/N) in UMI estimator | UMI estimator theory · Theoretical analysis | validity of reverse Fatou interchange | No clear effect |
| Use Renyi capacity (lambda != 1) instead of Shannon capacity CMI | axiomatic causal strength measure · Theoretical axioms | axiom satisfaction | No clear effect |
Closest papers by meaning
- On the sample complexity of conditional independence testing with Von Mises estimator with application to causal discovery (ICML 2024 · similarity 0.84)
- Extracting Rare Dependence Patterns via Adaptive Sample Reweighting (ICML 2025 · similarity 0.84)
- Optimal Experiments for Partial Causal Effect Identification (NeurIPS 2026 · similarity 0.84)
- Impossibility of Distribution-Free Predictive Inference for Individual Treatment Effects (NeurIPS 2026 · similarity 0.83)
- PAIR-CI: Calibrated Conditional Independence Testing for Causal Discovery with Incomplete Data (NeurIPS 2026 · similarity 0.83)
- Active Treatment Effect Estimation via Limited Samples (ICML 2025 · similarity 0.83)
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.
