Research map / Learning theory and algorithms / Sample complexity
Sensitivity-Aware Density Estimation in Multiple Dimensions
Aleix Boquet-Pujadas, Pol del Aguila Pla, Michael A. Unser
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 |
|---|---|---|---|
| adapt Lipschitz stepsize each iteration instead of fixed L(c1) (ablation) | RDS optimizer · synthetic 2D | convergence | Improved |
| use degree 1 splines instead of degree 0 (ablation) | RDS · synthetic densities | accuracy | Improved |
| use degree 3 splines instead of degree 1 (ablation) | RDS · synthetic densities | accuracy | No clear effect |
| fix lambda at optimum for 10^2 samples instead of adapting per sample count (ablation) | RDS · UGL density | MSE | No clear effect |
| use RDS instead of KDE and HE | density estimators · UGL 2D and 3D, sU and sS | MSE: order of magnitude improvement for >= 10^5 samples | Improved |
| use RDS instead of KDE and HE in 3D | density estimators · UGL 3D at 10^6 samples | MSE (dB): 15 dB | Improved |
| nonuniform sensitivity sS instead of uniform sU | RDS · 2D UGL | MSE (dB): 0.1 dB lost | Hurt |
| nonuniform sensitivity sS instead of uniform sU | KDE · 2D UGL | MSE (dB): 0.6 dB lost | Hurt |
| nonuniform sensitivity sS instead of uniform sU | HE · 2D UGL | MSE (dB): around 4 dB | Hurt |
| RDS instead of KDE for evaluation speed | RDS · 2D | evaluation time: 10^{-3} s vs 10^{-2} to 10^{2} s | Improved |
| RDS instead of HE and KDE for PET Derenzo FBP | density estimators · Derenzo phantom | MSE (dB): -184, -204 -> -211 | Improved |
| RDS instead of KDE for Digimouse FBP | density estimators · Digimouse | MSE (dB): -262 -> -269 | Improved |
Closest papers by meaning
- Adaptive Greedy Algorithm for Moderately Large Dimensions in Kernel Conditional Density Estimation (JMLR 2022 · similarity 0.86)
- Breaking the curse of dimensionality in structured density estimation (NeurIPS 2024 · similarity 0.85)
- Provable Posterior Sampling with Denoising Oracles via Tilted Transport (NeurIPS 2024 · similarity 0.83)
- Sobolev Space Regularised Pre Density Models (ICML 2024 · similarity 0.83)
- Posterior Sampling Based on Gradient Flows of the MMD with Negative Distance Kernel (ICLR 2024 · similarity 0.83)
- Stochastic Localization via Iterative Posterior Sampling (ICML 2024 · 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.
