RobustNeuralNetworks.jl: a Package for Machine Learning and Data-Driven Control with Certified Robustness
Open Access
- 13 February 2025
- journal article
- Published by The Open Journal in JuliaCon Proceedings
- Vol. 7 (68) , 163
Abstract
Neural networks are typically sensitive to small input perturbations, leading to unexpected or brittle behaviour. We present RobustNeuralNetworks.jl: a Julia package for neural network models that are constructed to naturally satisfy a set of userdefined robustness metrics. The package is based on the recently proposed Recurrent Equilibrium Network (REN) and Lipschitz-Bounded Deep Network (LBDN) model classes, and is designed to interface directly with Julia's most widely-used machine learning package, Flux.jl. We discuss the theory behind our model parameterization, give an overview of the package, and provide a tutorial demonstrating its use in image classification, reinforcement learning, and nonlinear state-observer design.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- On Robust Reinforcement Learning with Lipschitz-Bounded Policy Networks2025
- Learning Stable and Passive Neural Differential Equations2024
- Learning Over Contracting and Lipschitz Closed-Loops for Partially-Observed Nonlinear Systems2023
- Unconstrained Parametrization of Dissipative and Contracting Neural Ordinary Differential Equations2023
- Synthesis of Stabilizing Recurrent Equilibrium Network Controllers2022
- L2-Gain and Passivity Techniques in Nonlinear Control2017
- Deep Residual Learning for Image Recognition2016
- On Contraction Analysis for Non-linear SystemsAutomatica, 1998
- System analysis via integral quadratic constraintsIEEE Transactions on Automatic Control, 1997
- Stability of Discrete Linear Feedback SystemsIFAC Proceedings Volumes, 1975