arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.08822cs.CVcs.LGeess.IV
收藏

通过傅立叶掩模理解人工神经网络的鲁棒性和泛化

Understanding robustness and generalization of artificial neural networks through Fourier masks

Baylor College of Medicine(贝勒医学院);Rice University(莱斯大学)

Nikos Karantzas, Emma Besier, Josue Ortega Caro, Xaq Pitkow, Andreas S. Tolias, Ankit B. Patel, Fabio Anselmi

摘要:

尽管人工神经网络(ANN)在许多学科中取得了巨大的成功,但其计算的特征和关键属性(如泛化和鲁棒性)的起源仍然是一个悬而未决的问题。最近的文献表明,具有良好泛化性能的鲁棒网络往往偏向于处理图像中的低频。为了进一步探索频率偏差假设,我们开发了一种算法,使我们能够学习调制掩码,突出保持训练网络性能所需的基本输入频率。我们通过在输入频率中相对于这种调制的损耗中施加不变性来实现这一点。我们首先使用我们的方法来测试对抗训练或数据增强网络的低频偏好假设。我们的研究结果表明,对抗鲁棒网络确实表现出低频偏置,但我们发现这种偏置也依赖于频率空间中的方向。然而,这对于其他类型的数据增强不一定是正确的。我们的研究结果还表明,问题中的基本频率是有效的,用于实现泛化摆在首位。令人惊讶的是,通过这些调制面具看到的图像是不可识别的,并且类似于纹理样图案。

英文摘要:

Despite the enormous success of artificial neural networks (ANNs) in many disciplines, the characterization of their computations and the origin of key properties such as generalization and robustness remain open questions. Recent literature suggests that robust networks with good generalization properties tend to be biased towards processing low frequencies in images. To explore the frequency bias hypothesis further, we develop an algorithm that allows us to learn modulatory masks highlighting the essential input frequencies needed for preserving a trained network's performance. We achieve this by imposing invariance in the loss with respect to such modulations in the input frequencies. We first use our method to test the low-frequency preference hypothesis of adversarially trained or data-augmented networks. Our results suggest that adversarially robust networks indeed exhibit a low-frequency bias but we find this bias is also dependent on directions in frequency space. However, this is not necessarily true for other types of data augmentation. Our results also indicate that the essential frequencies in question are effectively the ones used to achieve generalization in the first place. Surprisingly, images seen through these modulatory masks are not recognizable and resemble texture-like patterns.

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。

↑