arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.03646cs.LGcs.AIcs.HC
收藏

通过将目标识别模型与人类脑电(EEG)联合训练实现对抗鲁棒性有限但一致的提升

Limited but consistent gains in adversarial robustness by co-training object recognition models with human EEG

Goethe University(歌德大学);The Hessian Center for Artificial Intelligence (hessian.AI)(黑森人工智能中心);Freie Universität Berlin(柏林自由大学);Frankfurt Institute for Advanced Studies (FIAS)(法兰克福高级研究所);Ernst Strüngmann Institute for Neuroscience(恩斯特·施特龙格曼神经科学研究所)

Manshan Guo, Bhavin Choksi, Sari Sadiya, Alessandro T. Gifford, Martina G. Vilas, Radoslaw M. Cichy, Gemma Roig

AI总结:

本研究通过训练ResNet50模型执行分类与EEG预测双任务,发现模型EEG预测准确率与对抗鲁棒性提升显著相关,效应有限但一致,顶枕区电极贡献最大,为相关后续研究开辟了方向。

摘要:

与人类视觉相比,人工神经网络(ANNs)仍相对容易受到对抗攻击的影响。为解决这一漏洞,研究人员已尝试将人类大脑的归纳偏置迁移到人工神经网络中,通常通过训练ANN表征使其匹配对应的生物表征来实现。以往研究依赖啮齿动物或灵长类动物通过侵入性技术采集的大脑数据,这些数据来自特定脑区,采集于非自然条件(麻醉动物)下,且所用刺激数据集缺乏多样性与自然性。\n在本研究中,我们探究了将模型表征与人类对丰富真实世界图像集的EEG响应对齐,是否能提升ANN的鲁棒性。具体而言,我们训练了以ResNet50为主干的模型,执行分类和EEG预测双重任务,并评估了模型的EEG预测准确率以及对抗攻击鲁棒性。我们观察到,网络的EEG预测准确率(通常在刺激出现后约100ms达到最高)与其对抗鲁棒性提升之间存在显著相关性。尽管效应量有限,但该效应在不同随机初始化下均一致存在,且对架构变体具有稳健性。我们进一步分离了单个EEG通道的数据,观察到顶枕区电极的贡献最强。人类EEG在此类任务中展现出的实用性,为未来拓展到更多样刺激条件下更大规模数据集、有望实现更强效应的研究开辟了道路。

英文摘要:

In contrast to human vision, artificial neural networks (ANNs) remain relatively susceptible to adversarial attacks. To address this vulnerability, efforts have been made to transfer inductive bias from human brains to ANNs, often by training the ANN representations to match their biological counterparts. Previous works relied on brain data acquired in rodents or primates using invasive techniques, from specific regions of the brain, under non-natural conditions (anesthetized animals), and with stimulus datasets lacking diversity and naturalness. In this work, we explored whether aligning model representations to human EEG responses to a rich set of real-world images increases robustness to ANNs. Specifically, we trained ResNet50-backbone models on a dual task of classification and EEG prediction; and evaluated their EEG prediction accuracy and robustness to adversarial attacks. We observed significant correlation between the networks' EEG prediction accuracy, often highest around 100 ms post stimulus onset, and their gains in adversarial robustness. Although effect size was limited, effects were consistent across different random initializations and robust for architectural variants. We further teased apart the data from individual EEG channels and observed strongest contribution from electrodes in the parieto-occipital regions. The demonstrated utility of human EEG for such tasks opens up avenues for future efforts that scale to larger datasets under diverse stimuli conditions with the promise of stronger effects.

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。

↑