/1D-LipCNNs

This repository contains the code for the paper "Lipschitz-bounded 1D convolutional neural networks using the Cayley transform and the controllability Gramian". We train Lipschitz-bounded 1D CNNs on the MIT-BIH heart arrythmia dataset.

Primary LanguageJulia

1D_LipCNNs

This code is based on the julia package RecurrentEquilibriumNetworks.jl which will be published soon. It further uses Flux and JuMP in combination with MOSEK.