Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Prototype Completion With Primitive Knowledge for Few-Shot Learning
Check 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 |
|---|---|---|---|
| Add prototype completion without fusion (ablation) | Mean-based prototype classifier · miniImagenet 5-way 1-shot | accuracy: 61.22 ± 0.84% -> 65.62 ± 0.79% | Helped |
| Add prototype completion without fusion (ablation) | Mean-based prototype classifier · miniImagenet 5-way 5-shot | accuracy: 78.72 ± 0.60% -> 75.32 ± 0.61% | Hurt |
| Add mean-based prototype fusion to completed prototypes (ablation) | ProtoComNet · miniImagenet 5-way 5-shot | accuracy: 75.32 ± 0.61% -> 79.70 ± 0.60% | Helped |
| Replace mean-based fusion with GaussFusion (ablation) | ProtoComNet with prototype fusion · miniImagenet 5-way 1-shot | accuracy: 70.14 ± 0.81% -> 73.13 ± 0.85% | Helped |
| Replace mean-based fusion with GaussFusion (ablation) | ProtoComNet with prototype fusion · miniImagenet 5-way 5-shot | accuracy: 79.70 ± 0.60% -> 82.06 ± 0.54% | Helped |
| Add GaussFusion to prototype completion (ablation) | ProtoComNet · miniImagenet 5-way 1-shot, primitive-knowledge noise γ=0.1 | accuracy: 59.28 % -> 71.80 % | Helped |
| Add GaussFusion to prototype completion (ablation) | ProtoComNet · miniImagenet 5-way 1-shot, primitive-knowledge noise γ=0.2 | accuracy: 55.39 % -> 70.77 % | Helped |
| Add GaussFusion to prototype completion (ablation) | ProtoComNet · miniImagenet 5-way 1-shot, primitive-knowledge noise γ=0.3 | accuracy: 51.97 % -> 69.97 % | Helped |
| Use prototype completion and GaussFusion instead of mean-based prototypes (ablation) | Few-shot classifier · miniImagenet 5-way 1-shot | accuracy: 61.22 ± 0.84% -> 73.13 ± 0.85% | Helped |
| Use prototype completion and GaussFusion instead of mean-based prototypes (ablation) | Few-shot classifier · miniImagenet 5-way 5-shot | accuracy: 78.72 ± 0.60% -> 82.06 ± 0.54% | Helped |
Closest papers by meaning
- A Closer Look at Prototype Classifier for Few-shot Image Classification (NeurIPS 2022 · similarity 0.91)
- MetaNODE: Prototype Optimization as a Neural ODE for Few-Shot Learning (AAAI 2022 · similarity 0.88)
- Compositional Prototypical Networks for Few-Shot Classification (AAAI 2023 · similarity 0.88)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.86)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.85)
- Mind the Gap Between Prototypes and Images in Cross-domain Finetuning (NeurIPS 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.