Research map / Probabilistic and scientific modeling / Variational and generative sampling
Adaptive Annealed Importance Sampling with Constant Rate Progress
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 a constant-rate, divergence-adaptive annealing schedule instead of a fixed heuristic schedule | AIS · 128D synthetic targets | absolute log Z estimation error: CR-AIS performs best on Student-T (0.69 ± 0.27); linear, sigmoidal, and exponential schedules perform best on the mixtur | Mixed |
| Use an interpolated CR-AIS schedule (ablation) | CR-AIS · 128D Gaussian mixture | absolute log Z estimation error: 308.82 ± 6.81 -> 228.54 ± 0.00 | Helped |
| Use an interpolated CR-AIS schedule (ablation) | CR-AIS · 128D Laplace | absolute log Z estimation error: 0.68 ± 0.31 -> 0.01 ± 0.00 | Helped |
| Use an interpolated CR-AIS schedule (ablation) | CR-AIS · 512D Laplace | absolute log Z estimation error: 8.75 ± 1.84 -> 7.93 ± 0.74 | Helped |
| Use an interpolated CR-AIS schedule (ablation) | CR-AIS · 512D Student-T | absolute log Z estimation error: 8.40 ± 1.28 -> 8.98 ± 0.00 | Hurt |
| Use CR-AIS instead of non-interpolated Adaptive AIS | AIS · 128D and 512D synthetic targets | computation complexity: about 3.5× more | Helped |
| Use CR-AIS instead of Adaptive AIS | AIS · Bayesian logistic regression on Pima and Sonar | target density evaluations for similar performance: × 4 fewer | Helped |
| Use a constant-rate schedule in place of ESS-based schedule adaptation | SMC · Four wide Gaussian mixture targets | mode coverage | Helped |
| Use the constant-rate schedule with geometric annealing | AIS · Inverse KL divergence along the annealing path | inverse KL divergence decrease rate | Helped |
| Use the alpha-power mean path with its derived constant-rate schedule | AIS · Alpha-divergence annealing | f-divergence decrease rate | Helped |
Closest papers by meaning
- Differentiable Annealed Importance Sampling Minimizes The Jensen-Shannon Divergence Between Initial and Target Distribution (ICML 2024 · similarity 0.89)
- Provable Benefit of Annealed Langevin Monte Carlo for Non-log-concave Sampling (ICLR 2025 · similarity 0.88)
- NETS: A Non-equilibrium Transport Sampler (ICML 2025 · similarity 0.88)
- Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics (ICLR 2025 · similarity 0.87)
- Non-equilibrium Annealed Adjoint Sampler (NeurIPS 2025 · similarity 0.87)
- Annealing Flow Generative Models Towards Sampling High-Dimensional and Multi-Modal Distributions (ICML 2025 · similarity 0.86)
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.