Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Image-Free Classifier Injection for Zero-Shot Classification
Anders Christensen, Massimiliano Mancini, A. Sophia Koepke, Ole Winther, Zeynep Akata
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 |
|---|---|---|---|
| replace L2 loss with cosine loss in descriptor-to-classifier MLP (ablation) | ICIS MLP base model · CUB GZSL | harmonic mean H: 0.0 -> 49.5 | Improved |
| replace L2 loss with cosine loss (ablation) | ICIS MLP base model · SUN ZSL | zero-shot accuracy: 49.7 -> 48.5 | Hurt |
| add within-space reconstruction losses on descriptors and classifiers (ablation) | ICIS · AWA2 ZSL | zero-shot accuracy: 50.9 -> 63.5 | Improved |
| add within-space reconstruction losses (ablation) | ICIS · SUN GZSL | harmonic mean H: 29.2 -> 35.4 | Improved |
| add across-space reconstruction (weights to descriptors) (ablation) | ICIS · CUB ZSL | zero-shot accuracy: 58.0 -> 60.1 | Improved |
| include unseen class descriptors during training (ablation) | ICIS · CUB ZSL | zero-shot accuracy: 60.1 -> 60.6 | No clear effect |
| include unseen descriptors during training (ablation) | ICIS · SUN GZSL | harmonic mean H: 33.9 -> 32.7 | Hurt |
| use ICIS instead of best baseline on image-free ZSL | ICIS · CUB, AWA2, SUN ZSL | zero-shot accuracy: +15.5% CUB and +9.2% AWA2 over SMO, +7.4% over ConSE on SUN | Improved |
| train ICIS from generic ImageNet classifier instead of seen-class classifier | ICIS · CUB and SUN ZSL (CLIP embeddings) | zero-shot accuracy: -22.5% CUB, -13.7% SUN | Hurt |
| train ICIS from ImageNet classifier with 1000 descriptor-classifier pairs | ICIS · AWA2 ZSL (CLIP embeddings) | zero-shot accuracy: up to 86.1 | Improved |
| use non-visual embeddings (Wiki2Vec, ConceptNet) with ImageNet-pretrained classifier on CUB | ICIS · CUB ZSL from ImageNet | zero-shot accuracy: 6.0 (wiki2vec), 10.2 (ConceptNet) | Hurt |
Closest papers by meaning
- Rethinking Zero-Shot Learning: A Conditional Visual Classification Perspective (ICCV 2019 · similarity 0.89)
- Learning Classifiers for Target Domain with Limited or No Labels (ICML 2019 · similarity 0.89)
- Classifier and Exemplar Synthesis for Zero-Shot Learning (IJCV 2019 · similarity 0.89)
- Synthesized Classifiers for Zero-Shot Learning (CVPR 2016 · similarity 0.88)
- Data-Free Generalized Zero-Shot Learning (AAAI 2024 · similarity 0.88)
- Class Semantic Attribute Perception Guided Zero-Shot Learning (AAAI 2025 · similarity 0.87)
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.
