aiqbench analysis
This research introduces a lightweight fault-detection scheme for Convolutional Neural Networks deployed on resource-constrained edge GPUs, addressing the critical need for reliability in safety-critical applications without significant latency penalties. By embedding dedicated filters that propagate checksums through inference layers, the method achieves high error detection rates with minimal overhead compared to existing matrix augmentation techniques. The findings are particularly relevant for engineers developing robust AI systems where soft errors could lead to unsafe decisions in embedded environments.
Carry-Through Checksum: A Lightweight Fault-Detection for CNN Inference at the Edge
arXiv:2609.16742v1 Announce Type: cross Abstract: Convolutional Neural Networks (CNNs) are increasingly deployed in safety-critical edge applications, where soft errors can silently corrupt inference outputs and lead to unsafe decisions. Such applications typically rely on resource-constrained embedded GPUs, requiring fault detection and mitigation techniques that add minimal compute, memory, and latency overhead while integrating seamlessly with the standard GPU inference pipeline. Existing algorithm-based fault tolerance techniques rely on matrix augmentation and per-operation checksum verification, imposing substantial overhead that is prohibitive for CNN inference on embedded GPUs. In this work, we propose carry-through checksum, a fundamentally new scheme for soft-error detection in CNN inference on embedded GPUs. The method embeds dedicated carry-through filters into the convolutional layers, which compute a checksum from the CNN's own operations and propagate it through inference, enabling end-to-end error detection with a single output verification. Experimental results on multiple CNN architectures show that the proposed method detects 95.86% and 86.56% of critical faults for FP32 and FP16, respectively, at almost no additional per-image overhead. Detected faults are mitigated through re-execution, incurring only 2.27% run-time overhead across the entire test set on an NVIDIA Jetson Orin NX GPU.
Abstract:Convolutional Neural Networks (CNNs) are increasingly deployed in safety-critical edge applications, where soft errors can silently corrupt inference outputs and lead to unsafe decisions. Such applications typically rely on resource-constrained embedded GPUs, requiring fault detection and mitigation techniques that add minimal compute, memory, and latency overhead while integrating seamlessly with the standard GPU inference pipeline. Existing algorithm-based fault tolerance techniques rely on matrix augmentation and per-operation checksum verification, imposing substantial overhead that is prohibitive for CNN inference on embedded GPUs. In this work, we propose carry-through checksum, a fundamentally new scheme for soft-error detection in CNN inference on embedded GPUs. The method embeds dedicated carry-through filters into the convolutional layers, which compute a checksum from the CNN's own operations and propagate it through inference, enabling end-to-end error detection with a single output verification. Experimental results on multiple CNN architectures show that the proposed method detects 95.86% and 86.56% of critical faults for FP32 and FP16, respectively, at almost no additional per-image overhead. Detected faults are mitigated through re-execution, incurring only 2.27% run-time overhead across the entire test set on an NVIDIA Jetson Orin NX GPU.
Accepted at ATS'26. 6 pages, 3 figs and 3 tables
Hardware Architecture (cs.AR); Machine Learning (cs.LG)
arXiv:2609.16742 [cs.AR]
(or arXiv:2609.16742v1 [cs.AR] for this version)
https://doi.org/10.48550/arXiv.2609.16742 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 292 words
Related coverage
Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
AWS · Sep 4, 2026
How Tata Elxsi detects industrial safety risks in seconds on AWS
AWS · Sep 22, 2026
Continuous Online Fault Detection for Mobile Robots via Adaptive Edge Models
arXiv · Sep 26, 2026
Reliability-Calibrated Edge-IoT Early Fault Warning for Rotating Machinery with a Physics-Guided Tiny-Mamba Transformer
arXiv · Jun 12, 2026