一句话：提出 Dr. DNA，通过神经元激活分布提取 SDC 签名，实现 DNN 推理中静默数据损坏的早期检测与缓解。
深度神经网络在自动驾驶等安全关键应用中广泛部署，但随着工艺缩放与硬件异构性增加，静默数据损坏（SDC）因其隐蔽性成为 DNN 可靠性的严重威胁。本文提出 Dr. DNA，一种通过神经元激活分布（DNA）来增强 DNN 系统可靠性的新方法。该方法从激活分布中形式化并提取一组独特的 SDC 签名，并据此在 DNN 推理早期检测与缓解 SDC。作者在 3 个视觉任务、5 个数据集、10 个模型和 4 种错误模型上进行了广泛评估。结果表明，Dr. DNA 在多数情况下达到 100% 的 SDC 检测率，平均检测率为 95%，所有情况下均超过 90%，相比基线提升 20%–70%；同时能以低于 1% 的内存开销和低于 2.5% 的延迟开销有效恢复模型性能，缓解 SDC 影响。
Deep neural networks (DNNs) have been widely-adopted in various safety-critical applications such as computer vision and autonomous driving. However, as technology scales and applications diversify, coupled with the increasing heterogeneity of underlying hardware architectures, silent data corruption (SDC) has been emerging as a pronouncing threat to the reliability of DNNs. Recent reports from industry hyperscalars underscore the difficulty in addressing SDC due to their "stealthy" nature and elusive manifestation. In this paper, we propose Dr. DNA, a novel approach to enhance the reliability of DNN systems by detecting and mitigating SDCs. Specifically, we formulate and extract a set of unique SDC signatures from the Distribution of Neuron Activations (DNA), based on which we propose early-stage detection and mitigation of SDCs during DNN inference. We perform an extensive evaluation across 3 vision tasks, 5 different datasets, and 10 different models, under 4 different error models. Results show that Dr. DNA achieves 100% SDC detection rate for most cases, 95% detection rate on average and >90% detection rate across all cases, representing 20% - 70% improvement over baselines. Dr. DNA can also mitigate the impact of SDCs by effectively recovering DNN model performance with <1% memory overhead and <2.5% latency overhead.
@inproceedings{MaLDCMSJ24,
title = {Dr. DNA: Combating Silent Data Corruptions in Deep Learning using Distribution of Neuron Activations},
author = {Dongning Ma and Fred Lin and Alban Desmaison and Joel Coburn and Daniel Moore and Sriram Sankar and Xun Jiao},
booktitle = {ASPLOS},
year = {2024},
doi = {10.1145/3620666.3651349}
}