Scholarlink

Nowadays, AI applications are becoming extremely popular in our everyday life as well as for the industry. Therefore, ensuring the trustworthiness of AI hardware is crucial, especially for edge applications in safety-critical systems. However, recent incidents involving major tech companies such as Google, Meta, and Alibaba have revealed that even cloud-based data center hardware can experience failures leading to Silent Data Corruptions (SDCs), also called Silent Data Errors (SDEs). This chapter delves into the implications of such failures on AI workloads, both during training and inference, and explores methodologies for efficiently detecting SDCs or SDEs through dedicated hardware monitors.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Trustworthy AI in the Cloud

  • 章节

  • 发表日期：

    2026年

查看全文
  • Alberto Bosio,
  • Ernesto Sanchez,
  • Arani Sinha,
  • Salvatore Pappalardo,
  • Annachiara Ruospo,
  • Vittorio Turco

摘要

Nowadays, AI applications are becoming extremely popular in our everyday life as well as for the industry. Therefore, ensuring the trustworthiness of AI hardware is crucial, especially for edge applications in safety-critical systems. However, recent incidents involving major tech companies such as Google, Meta, and Alibaba have revealed that even cloud-based data center hardware can experience failures leading to Silent Data Corruptions (SDCs), also called Silent Data Errors (SDEs). This chapter delves into the implications of such failures on AI workloads, both during training and inference, and explores methodologies for efficiently detecting SDCs or SDEs through dedicated hardware monitors.