Compute-in-Memory based Neural Network Accelerators for Safety-Critical Systems: Worst-Case Scenarios and Protections
Not on the map. Why?
Abstract (arXiv)
Emerging non-volatile memory (NVM)-based Computing-in-Memory (CiM) architectures show substantial promise in accelerating deep neural networks (DNNs) due to their exceptional energy efficiency. However, NVM devices are prone to device variations. Consequently, the actual DNN weights mapped to NVM devices can differ considerably from their targeted values, inducing significant performance degradation. Many existing solutions aim to optimize average performance amidst device variations, which is a suitable strategy for general-purpose conditions. However, the worst-case performance that is crucial for safety-critical applications is largely overlooked in current research. In this study, we define the problem of pinpointing the worst-case performance of CiM DNN accelerators affected by device variations. Additionally, we introduce a strategy to identify a specific pattern of the device value deviations in the complex, high-dimensional value deviation space, responsible for this worst-case outcome. Our findings reveal that even subtle device variations can precipitate a dramatic decline in DNN accuracy, posing risks for CiM-based platforms in supporting safety-critical applications. Notably, we observe that prevailing techniques to bolster average DNN performance in CiM accelerators fall short in enhancing worst-case scenarios. In light of this issue, we propose a novel worst-case-aware training technique named A-TRICE that efficiently combines adversarial training and noise-injection training with right-censored Gaussian noise to improve the DNN accuracy in the worst-case scenarios. Our experimental results demonstrate that A-TRICE improves the worst-case accuracy under device variations by up to 33%.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 17 of its 43 references that are in the map. First posted on arXiv 11 Dec 2023.
Builds on
17 of 43 references.
References · 17
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
- 2014Explaining and Harnessing Adversarial Examplescited by 7,287
- 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
- 2017Efficient Processing of Deep Neural Networks: A Tutorial and Surveycited by 427
- 2020Co-Exploration of Neural Architectures and Heterogeneous ASIC Accelerator Designs Targeting Multiple Taskscited by 43
- 2019Device-Circuit-Architecture Co-Exploration for Computing-in-Memory Neural Acceleratorscited by 32
- 2021Formalizing Generalization and Robustness of Neural Networks to Weight Perturbationscited by 22
- 2022The Larger The Fairer? Small Neural Networks Can Achieve Fairness for Edge Devicescited by 21
- 2021Uncertainty Modeling of Emerging Device-based Computing-in-Memory Neural Accelerators with Application to Neural Architecture Searchcited by 17
- 2022SWIM: Selective Write-Verify for Computing-in-Memory Neural Acceleratorscited by 16
- 2022Efficient Global Robustness Certification of Neural Networks via Interleaving Twin-Network Encodingcited by 15
5 more
- 2019When Single Event Upset Meets Deep Neural Networks: Observations, Explorations, and Remediescited by 14
- 2022Computing-In-Memory Neural Network Accelerators for Safety-Critical Systems: Can Small Device Variations Be Disastrous?cited by 12
- 2021RADARS: Memory Efficient Reinforcement Learning Aided Differentiable Neural Architecture Searchcited by 8
- 2022CorrectNet: Robustness Enhancement of Analog In-Memory Computing for Neural Networks by Error Suppression and Compensationcited by 8
- 2023Improving Realistic Worst-Case Performance of NVCiM DNN Accelerators through Training with Right-Censored Gaussian Noisecited by 8
Built on by
Citing papers per year: 2024: 1; 2025: 0; 2026: 1.
2 papers.