In a U-Net for segmentation, bit-flip damage concentrates in batch-norm gamma and bias parameters, and rewriting risky floating-point exponents can reduce error rates with no runtime overhead.
When single event upset meets deep neural networks: Observations, explorations, and remedies
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Evaluating Single Event Upsets in Deep Neural Networks for Semantic Segmentation: an embedded system perspective
In a U-Net for segmentation, bit-flip damage concentrates in batch-norm gamma and bias parameters, and rewriting risky floating-point exponents can reduce error rates with no runtime overhead.