Research map / Label-efficient and robust learning / Semi-supervised learning
Active Learning Through a Covering Lens
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 ProbCover instead of random sampling for low-budget selection | Active learning before FlexMatch · CIFAR-10, CIFAR-100, Tiny-ImageNet | test accuracy | Helped |
| Use ProbCover instead of alternative deep active-learning strategies | Low-budget image classification · CIFAR-10, CIFAR-100, Tiny-ImageNet, ImageNet | test accuracy | Helped |
| Use ProbCover instead of Coreset | Fully supervised ResNet-18 · CIFAR-10 low-budget regime | test accuracy | Helped |
| Use Coreset instead of ProbCover | Fully supervised ResNet-18 · CIFAR-10 high-budget regime | test accuracy | Mixed |
| Use ProbCover rather than random selection | Fully supervised ResNet-18 · CIFAR-10 mid-budget regime | test accuracy | Helped |
| Apply ProbCover with a shared random initial labeled pool (ablation) | ProbCover and TypiClust · CIFAR-10 and CIFAR-100 | test accuracy | Helped |
| Compute distances in RGB space instead of a semantic embedding (ablation) | ProbCover · Fully supervised image classification | test accuracy | Hurt |
| Use smaller δ as the query budget increases (ablation) | ProbCover · CIFAR-10 | accuracy difference from random selection | Helped |
| Use two-ball greedy selection instead of single-ball greedy selection (ablation) | ProbCover · Maximum coverage selection | accuracy | No clear effect |
| Select randomly among the five highest-degree nodes instead of selecting the highest-degree node (ablation) | ProbCover · Active-learning sample selection | performance | No clear effect |
| Use ProbCover rather than TypiClust | Fully supervised classification with SCAN features · CIFAR-10, budget of 10 examples | accuracy | Helped |
Closest papers by meaning
- Active Learning Guided by Efficient Surrogate Learners (AAAI 2024 · similarity 0.85)
- Composite Active Learning: Towards Multi-Domain Active Learning with Theoretical Guarantees (AAAI 2024 · similarity 0.85)
- A Cross-Domain Benchmark for Active Learning (NeurIPS 2024 · similarity 0.84)
- One-shot Active Learning Based on Lewis Weight Sampling for Multiple Deep Models (ICLR 2024 · similarity 0.84)
- Plug and Play Active Learning for Object Detection (CVPR 2024 · similarity 0.84)
- Active Prompt Learning in Vision Language Models (CVPR 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.