Research map / Learning theory and algorithms / Sample complexity
Mixing times of data-augmentation Gibbs samplers for high-dimensional probit regression
Filippo Ascolani, Giacomo Zanella
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 |
|---|---|---|---|
| increase g prior scale from g=1 to g=10 | DA Gibbs sampler · Imbalanced y=1, n/p=0.2, p=200 | upper bound on TV mixing time: 11 -> 65 | Hurt |
| increase isotropic prior variance from c=1 to c=10 | DA Gibbs sampler · Well-specified, n/p=3, p=200 | upper bound on TV mixing time: 11 -> 90 | Hurt |
| increase isotropic prior variance from c=1 to c=10 | Collapsed Gibbs sampler · Well-specified, n/p=3, p=200 | upper bound on TV mixing time: 31 -> 118 | Hurt |
| add an unscaled intercept column with imbalanced data | DA Gibbs sampler · Imbalanced y=1, n/p=3, p=200 | upper bound on TV mixing time: 9 -> 591 | Hurt |
| add an unscaled intercept column with well-specified data | DA Gibbs sampler · Well-specified, n/p=3, p=200 | upper bound on TV mixing time: 11 -> 17 | No clear effect |
| use collapsed instead of DA Gibbs sampler under g prior g=1 | Collapsed Gibbs sampler · Imbalanced y=1, n/p=3, p=200 | upper bound on TV mixing time: 6 -> 27 | Hurt |
| use collapsed instead of DA Gibbs sampler with c=10 prior | Collapsed Gibbs sampler · Well-specified, n/p=0.2, p=200 | upper bound on TV mixing time: 47 -> 21 | Improved |
| replace the DA sampler by the modified DA sampler that treats the intercept separately (ablation) | DA Gibbs sampler · Imbalanced data with intercept | upper bound on TV mixing time | Improved |
| scale prior variance as c=b/(1+r) with r=n/p | DA Gibbs sampler · Random design, theory | upper bound on TV mixing time | Improved |
Closest papers by meaning
- Fast Mixing Markov Chains for Strongly Rayleigh Measures, DPPs, and Constrained Sampling (NeurIPS 2016 · similarity 0.85)
- Fast Conditional Mixing of MCMC Algorithms for Non-log-concave Distributions (NeurIPS 2023 · similarity 0.85)
- Sharp Rates in Dependent Learning Theory: Avoiding Sample Size Deflation for the Square Loss (ICML 2024 · similarity 0.84)
- Active Treatment Effect Estimation via Limited Samples (ICML 2025 · similarity 0.84)
- Unified Convergence Analysis for Score-Based Diffusion Models with Deterministic Samplers (ICLR 2025 · similarity 0.83)
- Asymptotically Log-Optimal Bayes-Assisted Confidence Sequences for Bounded Means (NeurIPS 2026 · 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.
