Robust Methods for Deep-Learning Training

Editor’s notes: This article investigates the impact of hardware failures on deep-neuralnetwork training, discusses error detection and recovery strategies, and highlights key challenges for the future. —Jyotika Athavale, Synopsys, USA —Haralampos-G. Stratigopoulos, Sorbonne Université, CNRS, LIP6, France

• Publication year

  2025

• Venue

  IEEE design & test

• Publication date

  2025-06-01

• Fields of study

  Computer Science, Engineering

• Identifiers
  DOI 10.1109/MDAT.2024.3499878
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Understanding and Mitigating Hardware Failures in Deep Learning Training Systems

  2023cited by this paper
• Understanding Silent Data Corruptions in a Large Production CPU Population

  2023cited by this paper
• Understanding Permanent Hardware Failures in Deep Learning Training Accelerator Systems

  2023cited by this paper
• Testability and Dependability of AI Hardware: Survey, Trends, Challenges, and Perspectives

  2023cited by this paper
• Threats to Training: A Survey of Poisoning Attacks and Defenses on Machine Learning Systems

  2022cited by this paper
• Special Session: On the Reliability of Conventional and Quantum Neural Network Hardware

  2022cited by this paper
• Toward Understanding Deep Learning Framework Bugs

  2022influential reference
• Silent Data Corruptions at Scale

  2021cited by this paper
• CheckFreq: Frequent, Fine-Grained DNN Checkpointing

  2021cited by this paper
• Cores that don't count

  2021cited by this paper
• Failures Forecast in Monitoring Datacenter Infrastructure Through Machine Learning Techniques: A Systematic Review

  2021cited by this paper
• FIdelity: Efficient Resilience Analysis Framework for Deep Learning Accelerators

  2020cited by this paper
• A Low-cost Fault Corrector for Deep Neural Networks through Range Restriction

  2020cited by this paper
• A comprehensive study on deep learning bug characteristics

  2019cited by this paper
• Terminal Brain Damage: Exposing the Graceless Degradation in Deep Neural Networks Under Hardware Fault Attacks

  2019cited by this paper
• Ares: A framework for quantifying the resilience of deep neural networks

  2018cited by this paper
• Analyzing and mitigating the impact of permanent faults on a systolic array based neural network accelerator

  2018cited by this paper
• Understanding Error Propagation in Deep Learning Neural Network (DNN) Accelerators and Applications

  2017cited by this paper
• E-QED: Electrical Bug Localization During Post-silicon Validation Enabled by Quick Error Detection and Formal Methods

  2017cited by this paper
• CLEAR: Cross-layer exploration for architecting resilience: Combining hardware and software techniques to tolerate soft errors in processor cores

  2016cited by this paper
• Adam: A Method for Stochastic Optimization

  2014cited by this paper
• Quantitative evaluation of soft error injection techniques for robust system design

  2013cited by this paper

• No citing papers are available for this paper.