- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper presents CenWrite, a method specifically designed for cross-domain few-shot learning. This method centralizes the support set and uses the centralized activation values to learn LoRA. Furthermore, trainable residual adjustment is introduced to modify the classification boundaries. The experiment covered multiple datasets, including EuroSAT, PlantVillage, ISIC, NIH ChestX, as well as UHCS and NFFA, and provided a comparison with various adapter methods.
Critical strengths
1.The method is simple and easy to implement. 2.The experimental metrics are comprehensive, reporting Accuracy, Macro-F1, NLL, Brier, and ECE.
Critical weaknesses
1.In this paper, the optimization process is adjusted by removing the mean component of the support set features from the LoRA gradient. Although the authors provided derivations and ablation experiments, they failed to adequately explain why removing this component can improve the quality of predicted probabilities and how it specifically affects the optimization process. 2.CenWrite constructs feature centers by averaging all the supported set images and tokens. However, the rationale for this design is still unclear. This global averaging might mix the category-specific semantics with non-discriminative features. Since the feature centers directly guide the LoRA gradient modification,the authors should justify using a global mean over alternatives like class-wise means.
Questions for authors
1.Why does removing the gradient components related to the mean of the support set improve the quality of prediction probabilities? Please further analyze the underlying mechanism.
2.Why is the center of the support set calculated by averaging all the support set images and token positions? Have alternative strategies such as class centers been considered?
3.Why did CenWrite improve the probability metrics on DINOv3 but slightly reduce the classification accuracy and Macro-F1? Please further explain this trade-off and the reason for the inconsistent improvement between DINOv3 and DINOv2.
4.The paper primarily highlights CenWrite's improvements in probabilistic metrics such as NLL, Brier Score, and ECE. However, the experiments mainly compare against different parameter-efficient fine-tuning methods, lacking comparisons with simple probability calibration techniques.
5.The paper primarily adopts 5-way 1-shot and 5-way 5-shot settings, lacking a systematic evaluation across a wider range of class counts and support set sizes.