论文阅读 [TPAMI-2022] Performing Group Difference Testing on Graph Structured Data From GANs: Analysis a

论文阅读 [TPAMI-2022] Performing Group Difference Testing on Graph Structured Data From GANs: Analysis and Applications in Neuroimaging

论文搜索(studyai.com)

搜索论文: Performing Group Difference Testing on Graph Structured Data From GANs: Analysis and Applications in Neuroimaging

搜索论文: http://www.studyai.com/search/whole-site/?q=Performing+Group+Difference+Testing+on+Graph+Structured+Data+From+GANs:+Analysis+and+Applications+in+Neuroimaging

关键字(Keywords)

Gallium nitride; Statistical analysis; Diseases; Machine learning; Training data; Three-dimensional displays; Training; Generative adversarial network; graph theory; hypothesis testing; non-euclidean

机器学习; 机器视觉

生成对抗; 假设检验

摘要(Abstract)

Generative adversarial networks (GANs) have emerged as a powerful generative model in computer vision.

生成性对抗网络是计算机视觉中一种强大的生成模型。.

Given their impressive abilities in generating highly realistic images, they are also being used in novel ways in applications in the life sciences.

鉴于它们在生成高度逼真的图像方面令人印象深刻的能力，它们也以新颖的方式应用于生命科学中。.

This raises an interesting question when GANs are used in scientific or biomedical studies.

当GANs用于科学或生物医学研究时，这就提出了一个有趣的问题。.

Consider the setting where we are restricted to only using the samples from a trained GAN for downstream group difference analysis (and do not have direct access to the real data).

考虑这样的设置，我们只限于使用来自训练的GaN的样本来进行下游组差异分析（并且不直接访问真实数据）。.

Will we obtain similar conclusions? In this work, we explore if “generated” data, i.e., sampled from such GANs can be used for performing statistical group difference tests in cases versus controls studies, common across many scientific disciplines.

我们会得到类似的结论吗？在这项工作中，我们探索了“生成的”数据，即从这些GAN中取样的数据，是否可以用于在病例与对照研究中进行统计组差异测试，这在许多科学学科中都很常见。.

We provide a detailed analysis describing regimes where this may be feasible.

我们提供了详细的分析，描述了可能可行的制度。.

We complement the technical results with an empirical study focused on the analysis of cortical thickness on brain mesh surfaces in an Alzheimer’s disease dataset.

我们通过一项实证研究来补充技术结果，该研究侧重于分析阿尔茨海默病数据集中大脑网格表面的皮质厚度。.

To exploit the geometric nature of the data, we use simple ideas from spectral graph theory to show how adjustments to existing GANs can yield improvements.

为了利用数据的几何性质，我们使用谱图论的简单思想来展示如何调整现有的GAN来产生改进。.

We also give a generalization error bound by extending recent results on Neural Network Distance.

我们还通过扩展最近关于神经网络距离的结果给出了一个推广误差界。.

To our knowledge, our work offers the first analysis assessing whether the Null distribution in “healthy versus diseased subjects” type statistical testing using data generated from the GANs coincides with the one obtained from the same analysis with real data.

据我们所知，我们的工作提供了第一次分析，评估使用GANs生成的数据进行的“健康与疾病受试者”类型统计测试中的零分布是否与使用真实数据进行的相同分析中获得的零分布一致。.

The code is available at https://github.com/yyxiongzju/GLapGAN…

该代码可在https://github.com/yyxiongzju/GLapGAN…

作者(Authors)

[‘Tuan Q. Dinh’, ‘Yunyang Xiong’, ‘Zhichun Huang’, ‘Tien Vo’, ‘Akshay Mishra’, ‘Won Hwa Kim’, ‘Sathya N. Ravi’, ‘Vikas Singh’]