This repository implements the minimal finite covering problem and its application, and is the official source code for paper "Is Adversarial Training with Compressed Datasets Effective?" (Chen & Selvan. 2024). For more information, contact: Tong Chen (toch@di.ku.dk).
torch, numpy, scipy, gurobipy (license is required)
Finite covering with fixed number
sub_data, eta, sol = FC(data).gen_data(A, k=k, save=False)
Finite covering with fixed radius
sub_data, eta, sol = FC(data).gen_data(A, eta=eta, save=False)