Fast and Accurate Error Simulation for CNNs Against Soft Errors
Open Access
- 17 June 2022
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 72 (4) , 984-997
Abstract
The great quest for adopting AI-based computation for safety-/mission-critical applications motivates the interest towards methods for assessing the robustness of the application w.r.t. not only its training/tuning but also errors due to faults, in particular soft errors, affecting the underlying hardware. Two strategies exist: architecture-level fault injection and application-level functional error simulation. We present a framework for the reliability analysis of Convolutional Neural Networks (CNNs) via an error simulation engine that exploits a set of validated error models extracted from a detailed fault injection campaign. These error models are defined based on the corruption patterns of the output of the CNN operators induced by faults and bridge the gap between fault injection and error simulation, exploiting the advantages of both approaches. We compared our methodology against SASSIFI for the accuracy of functional error simulation w.r.t. fault injection, and against TensorFI in terms of speedup for the error simulation strategy. Experimental results show that our methodology achieves about 99% accuracy of the fault effects w.r.t. SASSIFI, and a speedup ranging from 44x up to 63x w.r.t. TensorFI, that only implements a limited set of error models.Keywords
All Related Versions
This publication has 32 references indexed in Scilit:
- Evaluation of Histogram of Oriented Gradients Soft Errors Criticality for Automotive ApplicationsACM Transactions on Architecture and Code Optimization, 2016
- Understanding Error Propagation in GPGPU Applications2016
- A Survey of Techniques for Approximate ComputingACM Computing Surveys, 2016
- Caffe2014
- Quantifying the Accuracy of High-Level Fault Injection Techniques for Hardware Faults2014
- GPU-Qin: A methodology for evaluating the error resilience of GPGPU applications2014
- Autonomous driving in urban environments: approaches, lessons and challengesPhilosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2010
- Characterization of soft errors caused by single event upsets in CMOS processesIEEE Transactions on Dependable and Secure Computing, 2004
- Single event upset at ground levelIEEE Transactions on Nuclear Science, 1996
- Backpropagation Applied to Handwritten Zip Code RecognitionNeural Computation, 1989