一句话：提出 SelfChecker，通过检查 DNN 内部层特征与最终预测的一致性，在部署中触发错误报警并给出备选预测。
深度神经网络在关键领域部署广泛，但即使高精度的 DNN 也会犯错，需在部署中快速检测。本文提出 DNN 自检过程并实现 SelfChecker：它监控 DNN 内部层特征，当特征与最终预测不一致时触发报警，并给出备选预测作为建议。在四个图像数据集和三个 DNN 模型上，SelfChecker 对错误预测的正确报警率为 60.56%，对正确预测的误报率为 2.04%，显著优于 SelfOracle、Dissector 和 ConfidNet；在自动驾驶场景中，对 DAVE-2 和 Chauffeur 也比 SelfOracle 触发更多正确报警且误报相当。
The widespread adoption of Deep Neural Networks (DNNs) in important domains raises questions about the trustworthiness of DNN outputs. Even a highly accurate DNN will make mistakes some of the time, and in settings like self-driving vehicles these mistakes must be quickly detected and properly dealt with in deployment. Just as our community has developed effective techniques and mechanisms to monitor and check programmed components, we believe it is now necessary to do the same for DNNs. In this paper we present DNN self-checking as a process by which internal DNN layer features are used to check DNN predictions. We detail SelfChecker, a self-checking system that monitors DNN outputs and triggers an alarm if the internal layer features of the model are inconsistent with the final prediction. SelfChecker also provides advice in the form of an alternative prediction. We evaluated SelfChecker on four popular image datasets and three DNN models and found that SelfChecker triggers correct alarms on 60.56% of wrong DNN predictions, and false alarms on 2.04% of correct DNN predictions. This is a substantial improvement over prior work (SelfOracle, Dissector, and ConfidNet). In experiments with self-driving car scenarios, SelfChecker triggers more correct alarms than SelfOracle for two DNN models (DAVE-2 and Chauffeur) with comparable false alarms. Our implementation is available as open source.
DOI 原文 · arXiv · PDF（开放获取） · DBLP
@inproceedings{XiaoBRSE0D21,
title = {Self-Checking Deep Neural Networks in Deployment},
author = {Yan Xiao and Ivan Beschastnikh and David S. Rosenblum and Changsheng Sun and Sebastian Elbaum and Yun Lin and Jin Song Dong},
booktitle = {ICSE},
year = {2021},
doi = {10.1109/icse43902.2021.00044}
}