mlsanityData to 7 Oct 2026

Winograd Convolution: A Perspective from Fault Tolerance

arXiv 2202.08675 ↗First posted 17 Feb 2022cs.LG cs.AR cs.PFcited by 6

Not on the map. Why?

Abstract (arXiv)

Winograd convolution is originally proposed to reduce the computing overhead by converting multiplication in neural network (NN) with addition via linear transformation. Other than the computing efficiency, we observe its great potential in improving NN fault tolerance and evaluate its fault tolerance comprehensively for the first time. Then, we explore the use of fault tolerance of winograd convolution for either fault-tolerant or energy-efficient NN processing. According to our experiments, winograd convolution can be utilized to reduce fault-tolerant design overhead by 27.49% or energy consumption by 7.19% without any accuracy loss compared to that without being aware of the fault tolerance

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 14 of its 48 references that are in the map. First posted on arXiv 17 Feb 2022.

Builds on

14 of 48 references.

References · 14

  1. 2015Fast Algorithms for Convolutional Neural Networkscited by 102
  2. 2021Robust Machine Learning Systems: Challenges, Current Trends, Perspectives, and the Road Aheadcited by 40
  3. 2020TensorFI: A Flexible Fault Injection Framework for TensorFlow Applicationscited by 21
  4. 2018Efficient Sparse-Winograd Convolutional Neural Networkscited by 20
  5. 2018ThUnderVolt: Enabling Aggressive Voltage Underscaling and Timing Error Resilience for Energy Efficient Deep Neural Network Acceleratorscited by 17
  6. 2020Searching for Winograd-aware Quantized Networkscited by 11
  7. 2021Arithmetic-Intensity-Guided Fault Tolerance for Neural Network Inference on GPUscited by 9
  8. 2018AxTrain: Hardware-Oriented Neural Network Training for Approximate Inferencecited by 8
  9. 2020FTT-NAS: Discovering Fault-Tolerant Convolutional Neural Architecturecited by 8
  10. 2019Training Modern Deep Neural Networks for Memory-Fault Robustnesscited by 6
  11. 2021R2F: A Remote Retraining Framework for AIoT Processors with Computing Errorscited by 6
  12. 2020DWM: A Decomposable Winograd Method for Convolution Accelerationcited by 5
2 more
  1. 2019CodeNet: Training Large Scale Neural Networks in Presence of Soft-Errorscited by 4
  2. 2020LANCE: Efficient Low-Precision Quantized Winograd Convolution for Neural Networks Based on Graphics Processing Unitscited by 4

Built on by

Cited by year

Citing papers per year: 2022: 2; 2023: 3; 2024: 1.

6 papers.

  1. 2022Fault-Tolerant Deep Learning: A Hierarchical Perspectivecited by 5
  2. 2023MRFI: An Open Source Multi-Resolution Fault Injection Framework for Neural Network Processingcited by 3
  3. 2022Statistical Modeling of Soft Error Influence on Neural Networkscited by 2
  4. 2023Exploring Winograd Convolution for Cost-effective Neural Network Fault Tolerancecited by 1
  5. 2023Cross-Layer Optimization for Fault-Tolerant Deep Learningcited by 0
  6. 2024Adaptive Soft Error Protection for Neural Network Processingcited by 0