Research map / Architectures and efficiency / RNNs and normalization
Prior-aware and Context-guided Group Sampling for Active Probabilistic Subsampling
Beomgu Kang, Hyunseok Seo
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 |
|---|---|---|---|
| add prior-aware deterministic sampling and group sampling to A-DPS | A-DPS · MNIST (1% sampling) | accuracy (%): 64.6 -> 68.8 | Improved |
| add prior-aware deterministic sampling and group sampling to A-DPS | A-DPS · CIFAR-10 (20% sampling) | accuracy (%): 68.3 -> 82.7 | Improved |
| add prior-aware deterministic sampling and group sampling to A-DPS | A-DPS · fastMRI knee 208x208 | PSNR (dB): 25.4 -> 25.9 | Improved |
| add prior-aware deterministic sampling and group sampling to A-DPS | A-DPS · fastMRI knee 368x640 | PSNR (dB): 30.2 -> 30.5 | Improved |
| add prior-aware deterministic sampling and group sampling to A-DPS | A-DPS · AeroRIT segmentation (5 bands) | mIOU: 0.5181 -> 0.6752 | Improved |
| increase active group size As from 5% to 20% without prior sampling (ablation) | A-DPS with group sampling · fastMRI knee 208x208 | NMSE: 0.0398 -> 0.0373 | Improved |
| use prior and active sampling (30%, 30%) instead of A-DPS (ablation) | PGA-DPS · fastMRI knee 208x208 | NMSE: 0.0398 -> 0.0359 | Improved |
| raise As beyond 15-20% (ablation) | PGA-DPS · fastMRI knee 208x208 | reconstruction metrics | No clear effect |
| use prior fraction Ps=80 instead of 20 with As=20 (ablation) | PGA-DPS · MNIST (1% sampling) | accuracy (%): 66.8 -> 69.1 | Improved |
| vary fixed temperature tau from 0.5 to 10.0 (ablation) | PGA-DPS · fastMRI knee 208x208 (M=26) | NMSE: 0.0354 -> 0.0349 | No clear effect |
| use A-DPS instead of DPS at higher sampling ratios | A-DPS · MNIST (8% sampling) | accuracy (%): 97.6 -> 97.2 | Hurt |
| use A-DPS instead of DPS | A-DPS · AeroRIT segmentation (5 bands) | mIOU: 0.6703 -> 0.5181 | Hurt |
Closest papers by meaning
- Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling (NeurIPS 2025 · similarity 0.84)
- Robust Sampling for Active Statistical Inference (NeurIPS 2025 · similarity 0.84)
- Scale-invariant Optimal Sampling for Rare-events Data and Sparse Models (NeurIPS 2024 · similarity 0.84)
- Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction (NeurIPS 2024 · similarity 0.84)
- Boundary Matters: A Bi-Level Active Finetuning Method (NeurIPS 2024 · similarity 0.84)
- State-of-art minibatches via novel DPP kernels: discretization, wavelets, and rough objectives (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.
