arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 1908.11230cs.LGcs.CR
收藏

克服迁移学习的误分类攻击

Defeating Misclassification Attacks Against Transfer Learning

Monash University(蒙纳士大学);CSIRO Data61(澳大利亚联邦科学与工业研究组织数据61);City University of Hong Kong(香港城市大学)

Bang Wu, Shuo Wang, Xingliang Yuan, Cong Wang, Carsten Rudolph, Xiangwen Yang

摘要:

迁移学习作为一种基于从预先训练的模型(教师模型)转移的知识来有效地生成新模型(学生模型)的技术而流行。然而,教师模型往往是公开的共享和重用,这不可避免地引入漏洞,触发严重的攻击迁移学习系统。在本文中,我们迈出了减轻迁移学习中最先进的误分类攻击的第一步。我们通过基于激活的网络修剪设计了一个蒸馏微分器,以削弱攻击的可转移性,同时保持准确性。我们采用一个集成结构,从不同的差异,以提高防御的鲁棒性。为了避免在推理过程中膨胀的集成规模,我们提出了一种两阶段防御,其中首先执行学生模型的推理以缩小待组装的候选微分器,然后只能选择其中一小部分固定数量的微分器来有效地验证干净或拒绝对抗性输入。我们对大型和小型图像识别任务的全面评估证实,我们仅使用5个微分器的学生模型对90%以上的对抗性输入免疫,准确性损失小于10%。我们的比较还表明,我们的设计优于以前的问题防御。

英文摘要:

Transfer learning is prevalent as a technique to efficiently generate new models (Student models) based on the knowledge transferred from a pre-trained model (Teacher model). However, Teacher models are often publicly available for sharing and reuse, which inevitably introduces vulnerability to trigger severe attacks against transfer learning systems. In this paper, we take a first step towards mitigating one of the most advanced misclassification attacks in transfer learning. We design a distilled differentiator via activation-based network pruning to enervate the attack transferability while retaining accuracy. We adopt an ensemble structure from variant differentiators to improve the defence robustness. To avoid the bloated ensemble size during inference, we propose a two-phase defence, in which inference from the Student model is firstly performed to narrow down the candidate differentiators to be assembled, and later only a small, fixed number of them can be chosen to validate clean or reject adversarial inputs effectively. Our comprehensive evaluations on both large and small image recognition tasks confirm that the Student models with our defence of only 5 differentiators are immune to over 90% of the adversarial inputs with an accuracy loss of less than 10%. Our comparison also demonstrates that our design outperforms prior problematic defences.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑