Verification of piecewise deep neural networks: a star set approach with zonotope pre-filter

DOI10.1007/s00165-021-00553-4OpenAlexW3198519141MaRDI QIDQ1982635

Xiao-Dong Yang, Taylor T. Johnson, Patrick Musau, Diego Manzanas Lopez, Luan Viet Nguyen, Neelanjana Pal, Weiming Xiang, Hoang-Dung Tran, Stanley Bak

Publication date: 14 September 2021

Published in: Formal Aspects of Computing (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1007/s00165-021-00553-4

zbMATH Keywords

robustnessneural networkssafetyformal verificationneural network verification

Mathematics Subject Classification ID

Computer science (68-XX)

Related Items (10)

Reachability analysis of deep ReLU neural networks using facet-vertex incidence ⋮ Stability analysis of complementarity systems with neural network controllers ⋮ Interval observer design of dynamical systems with neural networks ⋮ SOCKS: A Stochastic Optimal Control and Reachability Toolbox Using Kernel Methods ⋮ Verification of machine learning based cyber-physical systems: a comparative study ⋮ Fast BATLLNN: Fast Box Analysis of Two-Level Lattice Neural Networks ⋮ Quantitative Verification for Neural Networks using ProbStars ⋮ Verification of Recurrent Neural Networks with Star Reachability ⋮ Automatic Abstraction Refinement in Neural Network Verification using Sensitivity Analysis ⋮ BERN-NN: Tight Bound Propagation For Neural Networks Using Bernstein Polynomial Interval Arithmetic

Uses Software

• MPT
• MNIST
• DeepFool
• Reluplex
• NNV

Cites Work

• Parameterised verification for multi-agent systems
• Robustness verification of semantic segmentation neural networks using relaxed reachability
• Safety verification of deep neural networks
• Reluplex: an efficient SMT solver for verifying deep neural networks
• Simulation-equivalent reachability of large linear systems with inputs
• Verification of deep convolutional neural networks using ImageStars
• Improved geometric path enumeration for verifying ReLU neural networks
• Formal Verification of Piece-Wise Linear Feed-Forward Neural Networks
• Hybrid Systems: Computation and Control

This page was built for publication: Verification of piecewise deep neural networks: a star set approach with zonotope pre-filter