- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper proposes Patch Token Suppression, which learns a typographic score for each patch and moves representations with high scores toward a centroid estimated from patches assigned low scores in the same image. On synthetic ImageNet attacks, PTS improves object recovery while largely preserving clean accuracy. Overall, this is a solid paper that leans toward the accept side.
Critical strengths
-
The paper studies CLIP's robustness under typographic attacks, which has important practical significance;
-
The baseline methods are all state-of-the-art;
-
The dataset coverage is comprehensive, including multiple test datasets;
-
The reproducibility and ethics statements are well regulated. Random seeds, optimization settings, mask construction, and baseline reproduction details are provided.
Critical weaknesses
-
ImageNet appears multiple times in the main text but is not cited (it is only cited in the supplementary material), which is very unprofessional and shows that the authors' paper writing is not rigorous;
-
Only two models are evaluated, which is too few and may make it difficult to support the authors' claims;
-
The authors do not rigorously analyze or experimentally demonstrate the reasons for the drop in clean-sample accuracy;
-
The ablation experiments for some parameters are incomplete.
Questions for authors
See Weaknesses.