- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The core idea of the paper is to distill a universal perturbation that can be reused across different inputs from multiple successful individual attacks. The method can be summarized in three steps: (1) apply off-the-shelf individual attack methods to each sample and retain only successful results; (2) use the intermediate representations induced by these successful attacks as supervision targets; and (3) keep the model weights fixed while optimizing a single shared perturbation so that, for each input, it induces internal representations close to those produced by the corresponding successful individual attack.
Critical strengths
1.The proposed method is conceptually clear. It uses intermediate representations from successful individual attacks to supervise a universal perturbation, providing an effective way to reuse existing attack methods.
2.The framework has broad applicability and is experimentally evaluated in different application settings, namely image classification and LLM jailbreaking.
Critical weaknesses
-
I would like to see a numerical comparison of the training-set and test-set ASR achieved by the same final perturbation, to clearly demonstrate the generalization capability of the universal perturbation.
-
Why does increasing the cosine similarity between the intermediate representations produced by the teacher and student paths improve attack effectiveness? In other words, can cosine similarity capture differences in the key features that influence the network’s predictions? I would appreciate further theoretical explanation, or empirical evidence showing that cosine similarity is a comparatively suitable choice among commonly used metrics for measuring feature differences.
Questions for authors
Please refer to the questions raised in the weaknesses above.
Rebuttal to Reviewer 2vdW
We thank the reviewer for their thoughtful evaluation and for recognizing the strengths of our work. We address both concerns below and have incorporated the corresponding analyses and clarifications into the revised manuscript. We would be happy to provide further details or experiments.
Q1: Train vs Test Performance
We thank the reviewer for raising this question. We clarify that all results reported in the paper, except RQ1 (which examines optimization dynamics), are evaluated on held-out test sets. We additionally present cross-dataset generalization in Table 2, where UPD also leads.
The table below presents UPD ASR across data splits for LLMs, aggregated across models and datasets. There is no substantive difference between train and test ASRs, demonstrating generalization. Validation ASR is higher since it's used for best checkpoint selection. We have incorporated these results, alongside vision results into the revised manuscript (Appendix E.2).
Q2: Theoretical Justification for Similarity Function
We thank the reviewer for this question. The similarity function aims to steer representations into the failure region. At the penultimate layer, the key insight is that failure depends on representation direction rather than magnitude, and cosine similarity exactly respects this property. On the other hand, L2 penalizes differences in magnitude even when they do not affect predictions. This motivates cosine similarity at the penultimate layer, which UPD uses across both modalities on all models. Our existing ablation (Section 5.5) further shows that cosine similarity outperforms output-level cross-entropy, supporting the proposed objective.
For earlier layers, the failure-region geometry is unknown, but cosine similarity remains suited to the multi-layer objective, as its normalization prevents layers with larger activation norms from dominating the loss.
Formal justification. Let z be a penultimate representation, followed by a linear head W with no additive bias term, which yields the prediction argmaxj(Wz)j. For any α>0: argmaxj(W(αz))j=argmaxj(Wz)j, and therefore z∈F(x)⟺αz∈F(x). Membership in the failure region thus depends only on the direction of z, not on its norm.
For LLMs, the final unembedding matrix has no bias term, so the invariance holds exactly, and token rankings are retained. For vision models, the classification head often includes a bias term, so we evaluate the invariance empirically, and it is retained in 99.19% of cases across nine models and six UAP methods.
We added this discussion to the revised manuscript (Appendix B.3).