Detail publikace

Late Breaking Result: FPGA-Based Emulation and Fault Injection for CNN Inference Accelerators

MASÁR, F. MRÁZEK, V. SEKANINA, L.

Originální název

Late Breaking Result: FPGA-Based Emulation and Fault Injection for CNN Inference Accelerators

Typ

článek ve sborníku ve WoS nebo Scopus

Jazyk

angličtina

Originální abstrakt

A new field programmable gate array (FPGA)-based emulation platform is proposed to accelerate fault tolerance analysis of inference accelerators of convolutional neural networks (CNN). For a given CNN model, hardware accelerator architecture, and FT analysis target, an FPGA-based CNN implementation is generated (with the help of the Tengine framework), and fault injection logic is added. In our first case study, we report how the classification accuracy drop depends on the faults injected into multipliers used in Multiply-and-Accumulate Units of NVDLA inference accelerator executing ResNet-18 CNN. The FT analysis emulated on Zynq UltraScale+ SoC is an order of magnitude faster than software emulation.

Klíčová slova

Fault injection, hardware accelerator, convolutional neural network inference

Autoři

MASÁR, F.; MRÁZEK, V.; SEKANINA, L.

Vydáno

28. 2. 2025

Nakladatel

Institute of Electrical and Electronics Engineers

Místo

Lyon

ISBN

978-3-9826741-0-0

Kniha

2025 Design, Automation & Test in Europe Conference & Exhibition (DATE)

Strany od

1

Strany do

2

Strany počet

2

BibTex

@inproceedings{BUT193424,
  author="Filip {Masár} and Vojtěch {Mrázek} and Lukáš {Sekanina}",
  title="Late Breaking Result: FPGA-Based Emulation and Fault Injection for CNN Inference Accelerators",
  booktitle="2025 Design, Automation & Test in Europe Conference & Exhibition (DATE)",
  year="2025",
  pages="1--2",
  publisher="Institute of Electrical and Electronics Engineers",
  address="Lyon",
  doi="10.23919/DATE64628.2025.10992992",
  isbn="978-3-9826741-0-0"
}