This paper aims to enhance the transferability of adversarial samples in targeted attacks, where attack success rates remain comparatively low. To achieve this objective, we propose two distinct techniques for improving the targeted transferability from the loss and feature aspects. First, in previous approaches, logit calibrations used in targeted attacks primarily focus on the logit margin between the targeted class and the untargeted classes among samples, neglecting the standard deviation of the logit. In contrast, we introduce a new normalized logit calibration method that jointly considers the logit margin and the standard deviation of logits. This approach effectively calibrates the logits, enhancing the targeted transferability. Second, previous studies have demonstrated that mixing the features of clean samples during optimization can significantly increase transferability. Building upon this, we further investigate a truncated feature mixing method to reduce the impact of the source training model, resulting in additional improvements. The truncated feature is determined by removing the Rank-1 feature associated with the largest singular value decomposed from the high-level convolutional layers of the clean sample. Extensive experiments conducted on the ImageNet-Compatible and CIFAR-10 datasets demonstrate the individual and mutual benefits of our proposed two components, which outperform the state-of-the-art methods by a large margin in black-box targeted attacks.


    Access

    Download


    Export, share and cite



    Title :

    Improving Transferable Targeted Adversarial Attack via Normalized Logit Calibration and Truncated Feature Mixing


    Contributors:
    Weng, Juanjuan (author) / Luo, Zhiming (author) / Li, Shaozi (author)

    Publication date :

    2024


    Type of media :

    Preprint


    Type of material :

    Electronic Resource


    Language :

    English




    Adversarial Logit Pairing

    Kannan, Harini / Kurakin, Alexey / Goodfellow, Ian | ArXiv | 2018

    Free access

    Improving VAEs' Robustness to Adversarial Attack

    Willetts, Matthew / Camuto, Alexander / Rainforth, Tom et al. | ArXiv | 2019

    Free access

    LoCa: Logit Calibration for Knowledge Distillation

    Yang, Runming / Wu, Taiqiang / Yang, Yujiu | ArXiv | 2024

    Free access

    Normalized truncated Levy walk applied to flexible pavement performance

    Mariani, Maria Cristina | Online Contents | 2012