arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.08865cs.CVcs.LGeess.IVstat.ML

数据集内在属性对泛化的影响：揭示自然图像与医学图像之间的学习差异

The Effect of Intrinsic Dataset Properties on Generalization: Unraveling Learning Differences Between Natural and Medical Images

Duke University（杜克大学）

Nicholas Konz, Maciej A. Mazurowski

阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文通过提出标签锐度度量并建立关于数据内在维度的泛化缩放定律，解释了自然图像与医学图像间的泛化差异，并揭示了医学图像模型对抗鲁棒性更弱的原因。

摘要：

本文研究了神经网络在不同成像域中学习方式的差异，这些差异在将计算机视觉技术从自然图像域迁移到医学图像等专业领域时常常被忽视。近期研究发现，训练好的网络的泛化误差通常随其训练集的内在维度（$d_{data}$）增加而增大。然而，这一关系的陡峭程度在医学（放射学）图像域和自然图像域之间存在显著差异，目前尚无理论解释。我们通过建立并实证验证关于 $d_{data}$ 的泛化缩放定律来填补这一知识空白，并提出所考虑的两个域之间的显著缩放差异可能至少部分归因于医学影像数据集具有更高的内在“标签锐度”（$K_\mathcal{F}$）——这是我们提出的一个度量指标。接下来，我们展示了测量训练集标签锐度的另一个益处：它与训练模型的对抗鲁棒性呈负相关，这显著导致医学图像模型对对抗攻击具有更高的脆弱性。最后，我们将 $d_{data}$ 形式化方法扩展到相关的学习表示内在维度（$d_{repr}$）度量，推导出关于 $d_{repr}$ 的泛化缩放定律，并证明 $d_{data}$ 是 $d_{repr}$ 的上界。我们的理论结果得到了充分实验的支持，实验涉及六种模型和十一个自然与医学影像数据集，并覆盖了一系列训练集规模。我们的发现为理解数据集内在属性对深度神经网络泛化、表示学习和鲁棒性的影响提供了见解。代码链接：https://github.com/mazurowski-lab/intrinsic-properties

英文摘要：

This paper investigates discrepancies in how neural networks learn from different imaging domains, which are commonly overlooked when adopting computer vision techniques from the domain of natural images to other specialized domains such as medical images. Recent works have found that the generalization error of a trained network typically increases with the intrinsic dimension ($d_{data}$) of its training set. Yet, the steepness of this relationship varies significantly between medical (radiological) and natural imaging domains, with no existing theoretical explanation. We address this gap in knowledge by establishing and empirically validating a generalization scaling law with respect to $d_{data}$, and propose that the substantial scaling discrepancy between the two considered domains may be at least partially attributed to the higher intrinsic ``label sharpness'' ($K_\mathcal{F}$) of medical imaging datasets, a metric which we propose. Next, we demonstrate an additional benefit of measuring the label sharpness of a training set: it is negatively correlated with the trained model's adversarial robustness, which notably leads to models for medical images having a substantially higher vulnerability to adversarial attack. Finally, we extend our $d_{data}$ formalism to the related metric of learned representation intrinsic dimension ($d_{repr}$), derive a generalization scaling law with respect to $d_{repr}$, and show that $d_{data}$ serves as an upper bound for $d_{repr}$. Our theoretical results are supported by thorough experiments with six models and eleven natural and medical imaging datasets over a range of training set sizes. Our findings offer insights into the influence of intrinsic dataset properties on generalization, representation learning, and robustness in deep neural networks. Code link: https://github.com/mazurowski-lab/intrinsic-properties

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。

↑