Research map / Continual learning / Class-incremental
Few-Shot Continual Active Learning by a Robot
Ali Ayub, Carter Fendley
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 |
|---|---|---|---|
| Use GBCL rather than random sampling (ablation) | CORe-50 FoCAL · CORe-50 | accuracy | Improved |
| Combine predictive entropy and viewpoint consistency rather than use either acquisition function alone (ablation) | GBCL active learning · Pepper dataset | accuracy: only 0.07% | Improved |
| Set the entropy weight to 0.7 rather than use either acquisition function alone (ablation) | GBCL active learning · Pepper dataset | accuracy: 0.7; only 0.07% | Improved |
| Use a probability threshold near 0 rather than the selected threshold (ablation) | GBCL GMM clustering · Pepper dataset | accuracy: ∼18% | Hurt |
| Increase the GMM probability threshold to 0.15 (ablation) | GBCL GMM clustering · Pepper dataset | average incremental accuracy: P=0.15; 41 total clusters | Improved |
| Increase the GMM probability threshold to 0.2 (ablation) | GBCL GMM clustering · Pepper dataset | accuracy: P=0.2; 83 total clusters | Improved |
| Increase the GMM probability threshold to 0.7 (ablation) | GBCL GMM clustering · Pepper dataset | average incremental accuracy: P=0.7; 379 clusters; 4.5 times | Improved |
| Use a probability threshold near 1, recruiting many clusters with few features each (ablation) | GBCL GMM clustering · Pepper dataset | storage use | Hurt |
| Use GBCL rather than FLB | CORe-50 FoCAL · CORe-50, 400 increments | accuracy: ∼89% vs ∼88%; 1% lower | Hurt |
| Use GBCL rather than CBCL | CORe-50 FoCAL · CORe-50 | accuracy | Improved |
| Use GMM-based active learning rather than random sampling (ablation) | GBCL · CORe-50, early increments | accuracy | Improved |
| Use a fixed feature extractor rather than adapt the representation during incremental learning | GBCL vs TOPIC · CIFAR-100 FSCIL | accuracy: ∼8% | Mixed |
Closest papers by meaning
- Re-evaluating Continual Learning with Few-Shot Adaptation (NeurIPS 2026 · similarity 0.88)
- FLAR: A Unified Prototype Framework for Few-Sample Lifelong Active Recognition (ICCV 2021 · similarity 0.87)
- Active Learning for Continual Learning: Keeping the Past Alive in the Present (ICLR 2025 · similarity 0.86)
- Stabilizing Zero-Shot Prediction: A Novel Antidote to Forgetting in Continual Vision-Language Tasks (NeurIPS 2024 · similarity 0.85)
- Online Continual Learning for Interactive Instruction Following Agents (ICLR 2024 · similarity 0.85)
- Enhancing Visual Continual Learning with Language-Guided Supervision (CVPR 2024 · similarity 0.85)
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.
