Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Improving Zero-Shot Generalization for CLIP with Variational Adapter
Ziqian Lu, Fengli Shen, Mushui Liu, Yunlong Yu, Xi Li
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 |
|---|---|---|---|
| add classification loss Lcls to Ldis (ablation) | PVA · CUB GZSL | harmonic mean H: 22.5 -> 66.1 | Improved |
| add cross-domain reconstruction loss Lcr (ablation) | PVA · AwA2 GZSL | harmonic mean H: 86.6 -> 90.3 | Improved |
| add refinement with CLIP global features (residual, epsilon 0.9) (ablation) | PVA · CUB GZSL | harmonic mean H: 69.6 -> 71.4 | Improved |
| add refinement with CLIP global features (ablation) | PVA · AwA2 GZSL | harmonic mean H: 90.3 -> 90.7 | No clear effect |
| replace variational adapter with MLP adapter plus data separation (ablation) | PVA · AwA2 GZSL | harmonic mean H: 90.7 -> 79.1 | Hurt |
| add data separation to MLP adapter (ablation) | MLP adapter · AwA2 GZSL | harmonic mean H: 85.0 -> 79.1 | Hurt |
| add data separation to MLP adapter (ablation) | MLP adapter · CUB GZSL | harmonic mean H: 57.0 -> 66.6 | Improved |
| replace Ldis-only domain detector with full losses (ablation) | PVA domain detection · CUB | FPR at 95% TPR: 77.1 -> 2.6 | Improved |
| use CLIP features in generative method TF-VAEGAN | TF-VAEGAN · AwA2 GZSL | harmonic mean H | No clear effect |
| increase threshold ratio tau to 0.99 or above (ablation) | PVA · AwA2 GZSL | novel-class accuracy N | Hurt |
| fine-tune CLIP with CoOP prompt learning | CLIP · AwA2 GZSL | novel-class accuracy N: 86.6 -> 72.7 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- PSVMA — PSVMA+: Exploring Multi-Granularity Semantic-Visual Adaption for Generalized Zero-Shot Learning (TPAMI 2024) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Test-Time Adaptation with CLIP Reward for Zero-Shot Generalization in Vision-Language Models (ICLR 2024 · similarity 0.90)
- Conformal Prediction for Zero-Shot Models (CVPR 2025 · similarity 0.89)
- On the Test-Time Zero-Shot Generalization of Vision-Language Models: Do We Really Need Prompt Learning? (CVPR 2024 · similarity 0.89)
- TPR: Topology-Preserving Reservoirs for Generalized Zero-Shot Learning (NeurIPS 2024 · similarity 0.89)
- MMA: Multi-Modal Adapter for Vision-Language Models (CVPR 2024 · similarity 0.88)
- A Hard-to-Beat Baseline for Training-free CLIP-based Adaptation (ICLR 2024 · similarity 0.88)
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.
