Research map / Learning theory and algorithms / Sample complexity
Approximate Inference via Weighted Rademacher Complexity
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 |
|---|---|---|---|
| Use weighted Rademacher perturbations instead of Gumbel perturbations | propositional model counting · log-1 and tire-1 | upper bound on ln(model count): 438.0 (46.2) vs 426.5 (43.0); 198.5 (17.6) vs 249.6 (23.7) | Mixed |
| Use weighted Rademacher perturbations instead of Gumbel perturbations | propositional model counting · ra, sat-grid-pbl-0010, and c432 | upper bound on ln(model count): 856.7 (0.0) vs 1100.9 (45.7); 76.2 (0.0) vs 176.3 (13.6); 135.3 (1.0) vs 212.6 (18.2) | Helped |
| Use weighted Rademacher perturbations instead of Gumbel perturbations | propositional model counting · log-1, tire-1, ra, sat-grid-pbl-0010, and c432 | lower bound on ln(model count): 0.5 (0.6) vs -0.3 (0.0); 0.3 (0.4) vs -0.5 (0.1); 184.1 (10.1) vs 0.3 (0.0); 7.6 (2.2) vs -0.5 (0.1); 1.4 (0.8) vs -0.5 | Helped |
| Use weighted Rademacher perturbations instead of Gumbel perturbations | spin glass partition-function bounding · 7x7 spin glass | upper bound on log partition function: null | Helped |
| Increase the number of perturbation samples | spin glass partition-function bounding · 7x7 spin glass | upper bound on log partition function: k=5; k ≈ 15 | Mixed |
| Use weighted Rademacher perturbations instead of Gumbel perturbations | propositional model counting · #SAT instances | computation time: over 2 orders of magnitude | Helped |
Closest papers by meaning
- On Contraction of Sequential and Offset Rademacher Complexities (ICML 2026 · similarity 0.90)
- Hypothesis Set Stability and Generalization (NeurIPS 2019 · similarity 0.90)
- On the Rademacher Complexity of Graph Neural Networks: Unifying Expressivity and Geometry (NeurIPS 2026 · similarity 0.86)
- Data-dependent Sample Complexity of Deep Neural Networks via Lipschitz Augmentation (NeurIPS 2019 · similarity 0.86)
- Non-Asymptotic Length Generalization (ICML 2025 · similarity 0.85)
- An Approximate Skolem Function Counter (AAAI 2024 · similarity 0.84)
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.