arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.08985cs.LG
收藏

基于丢弃贝叶斯神经网络的分布外数据检测

Out of Distribution Data Detection Using Dropout Bayesian Neural Networks

Laboratory for Physical Sciences(物理科学实验室);Booz Allen Hamilton(博思艾伦汉密尔顿);University of Maryland, Baltimore County(马里兰大学巴尔的摩县分校)

Andre T. Nguyen, Fred Lu, Gary Lopez Munoz, Edward Raff, Charles Nicholas, James Holt

摘要:

我们探讨了实用的信息包含在一个辍学的贝叶斯神经网络(BNN)的任务检测出分布(OOD)的数据。我们首先展示了以前的尝试如何利用由辍学BNN的中间层引起的随机嵌入,由于使用的距离度量而失败。我们介绍了一种替代方法来测量嵌入的不确定性,证明其使用理论上,并演示了如何将嵌入的不确定性提高OOD数据识别在三个任务:图像分类,语言分类和恶意软件检测。

英文摘要:

We explore the utility of information contained within a dropout based Bayesian neural network (BNN) for the task of detecting out of distribution (OOD) data. We first show how previous attempts to leverage the randomized embeddings induced by the intermediate layers of a dropout BNN can fail due to the distance metric used. We introduce an alternative approach to measuring embedding uncertainty, justify its use theoretically, and demonstrate how incorporating embedding uncertainty improves OOD data identification across three tasks: image classification, language classification, and malware detection.

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。

↑