arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2207.04075cs.LG
收藏

Out of Line模型:基于分布偏移鲁棒性的傅立叶透镜

Models Out of Line: A Fourier Lens on Distribution Shift Robustness

UC Berkeley(加州大学伯克利分校);Lawrence Livermore National Laboratory(劳伦斯利弗莫尔国家实验室)

Sara Fridovich-Keil, Brian R. Bartoldson, James Diffenderfer, Bhavya Kailkhura, Peer-Timo Bremer

摘要:

提高深度神经网络(DNN)对分布外(OOD)数据的准确性对于在现实世界应用中接受深度学习(DL)至关重要。已经观察到,分布(ID)与OOD数据的准确性遵循线性趋势,并且优于该基线的模型非常罕见(并被称为“有效稳健”)。最近,已经开发了一些有前途的方法来提高OOD的鲁棒性:模型修剪,数据增强,以及集成或zero-shot评估大型预训练模型。然而,仍然没有明确的理解OOD数据的条件和模型属性,需要观察有效的鲁棒性。我们通过对各种方法进行全面的实证研究来解决这个问题,这些方法已知会对CIFAR-10和ImageNet的广泛自然和合成分布偏移产生影响。特别是,我们认为“有效的鲁棒性难题”,通过傅立叶透镜,并询问光谱特性的模型和OOD数据如何影响相应的有效的鲁棒性。我们发现这个傅立叶透镜提供了一些见解,为什么某些强大的模型,特别是那些从CLIP家庭,实现OOD的鲁棒性。然而,我们的分析也清楚地表明,没有任何已知的度量标准是OOD鲁棒性的最佳解释(甚至是强有力的解释)。因此,为了帮助未来对OOD难题的研究,我们通过引入一组具有不同OOD鲁棒性水平的预训练模型-RobustNets-来解决具有有效鲁棒性的公开模型中的差距。

英文摘要:

Improving the accuracy of deep neural networks (DNNs) on out-of-distribution (OOD) data is critical to an acceptance of deep learning (DL) in real world applications. It has been observed that accuracies on in-distribution (ID) versus OOD data follow a linear trend and models that outperform this baseline are exceptionally rare (and referred to as "effectively robust"). Recently, some promising approaches have been developed to improve OOD robustness: model pruning, data augmentation, and ensembling or zero-shot evaluating large pretrained models. However, there still is no clear understanding of the conditions on OOD data and model properties that are required to observe effective robustness. We approach this issue by conducting a comprehensive empirical study of diverse approaches that are known to impact OOD robustness on a broad range of natural and synthetic distribution shifts of CIFAR-10 and ImageNet. In particular, we view the "effective robustness puzzle" through a Fourier lens and ask how spectral properties of both models and OOD data influence the corresponding effective robustness. We find this Fourier lens offers some insight into why certain robust models, particularly those from the CLIP family, achieve OOD robustness. However, our analysis also makes clear that no known metric is consistently the best explanation (or even a strong explanation) of OOD robustness. Thus, to aid future research into the OOD puzzle, we address the gap in publicly-available models with effective robustness by introducing a set of pretrained models--RobustNets--with varying levels of OOD robustness.

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。

↑