New framework statistically analyzes Vision Transformer reliability

Researchers have developed SENTRY, a statistical framework to analyze the reliability of Vision Transformers (ViTs) against soft errors. This method uses finite-population sampling theory to provide formal reliability guarantees, significantly reducing the cost of fault injection campaigns. Evaluations on ViT-Tiny and ViT-Small models revealed that while few bit-flips cause failure, those that do lead to drastic accuracy drops, often localized in normalization layers and specific floating-point bits. AI

IMPACT Provides a cost-effective method to ensure the reliability of vision models in critical applications.

RANK_REASON This is a research paper detailing a new methodology for analyzing model reliability. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• SENTRY
• Vision Transformers
• ViT-Small
• ViT-Tiny

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →