Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
XtarNet: Learning to Extract Task-Adaptive Representation for Incremental Few-Shot Learning
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 XtarNet with TapNet instead of Attention Attractor Networks | Incremental few-shot learning · miniImageNet 64+5-way, 1-shot | accuracy: 54.95 ± 0.30% -> 55.28 ± 0.33% | Helped |
| Use XtarNet with TapNet instead of Attention Attractor Networks | Incremental few-shot learning · miniImageNet 64+5-way, 5-shot | accuracy: 63.04 ± 0.30% -> 66.86 ± 0.31% | Helped |
| Use XtarNet with TapNet instead of Attention Attractor Networks | Incremental few-shot learning · tieredImageNet 200+5-way, 1-shot | accuracy: 56.11 ± 0.33% -> 61.37 ± 0.36% | Helped |
| Use XtarNet with TapNet instead of Attention Attractor Networks | Incremental few-shot learning · tieredImageNet 200+5-way, 5-shot | accuracy: 65.52 ± 0.31% -> 69.58 ± 0.32% | Helped |
| Apply TAR to Imprint | Imprint · miniImageNet 64+5-way, 5-shot | accuracy: 46.34 ± 0.54% -> 56.66 ± 0.51% | Helped |
| Apply TAR to Imprint | Imprint · tieredImageNet 200+5-way, 5-shot | accuracy: 53.87 ± 0.48% -> 65.56 ± 0.50% | Helped |
| Add MetaCNN to Imprint (ablation) | Imprint · tieredImageNet, 5-shot | accuracy: 53.87 ± 0.48% -> 63.15 ± 0.49% | Helped |
| Add MergeNet to MetaCNN + Imprint (ablation) | MetaCNN + Imprint · tieredImageNet, 5-shot | accuracy: 63.15 ± 0.49% -> 64.62 ± 0.51% | Helped |
| Add TconNet to MetaCNN + MergeNet + Imprint (ablation) | MetaCNN + MergeNet + Imprint · tieredImageNet, 5-shot | accuracy: 64.62 ± 0.51% -> 65.56 ± 0.50% | Helped |
| Combine base and MetaCNN novel features using task-adaptive mixing | Imprint · tieredImageNet, 5-shot, base and novel classes | SSE: 10.65 -> 4.94; 3.96 -> 2.38 | Helped |
Closest papers by meaning
- TapNet: Neural Network Augmented with Task-Adaptive Projection for Few-Shot Learning (ICML 2019 · similarity 0.87)
- Few-Shot Hybrid Incremental Learning:Continually Learning under Data Scarcity and Task Uncertainty (CVPR 2026 · similarity 0.87)
- Finding Task-Relevant Features for Few-Shot Learning by Category Traversal (CVPR 2019 · similarity 0.87)
- Tripartite Weight-Space Ensemble for Few-Shot Class-Incremental Learning (CVPR 2025 · similarity 0.86)
- Few-Shot Incremental Learning With Continually Evolved Classifiers (CVPR 2021 · similarity 0.86)
- Adaptive Decision Boundary for Few-Shot Class-Incremental Learning (AAAI 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.