Scholarlink

PyRAT, a tool based on abstract interpretation to verify the safety and robustness of neural networks, is participating in VNN-Comp for the third time in a row. PyRAT uses multiple abstractions to find the reachable states of a neural network, starting from its input and propagating it through the layers in a fast and accurate analysis. It has been applied on public benchmarks as well as industrial use-cases.

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

PyRAT: Verifying Neural Networks with Abstract Interpretation (Competition Contribution)

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Augustin Lemesle,
  • Julien Lehmann,
  • Tristan Le Gall,
  • Zakaria Chihani

摘要

PyRAT, a tool based on abstract interpretation to verify the safety and robustness of neural networks, is participating in VNN-Comp for the third time in a row. PyRAT uses multiple abstractions to find the reachable states of a neural network, starting from its input and propagating it through the layers in a fast and accurate analysis. It has been applied on public benchmarks as well as industrial use-cases.