Fig. 8: The prediction results for subsystem entropy \(S({\rho }_{A})=-\log (\,\text{tr}\,{\rho }_{A}^{2})\), with ρA = [1], [12], and [123], were obtained for two image adversarial situations.
From: Direct entanglement detection of quantum systems using machine learning

a The predicted entropies of the separable Hamiltonians ground states. b The predicted entropies of the fully separable states. For each adversarial situation, we tested M = 500 samples. The red points represent the actual entropy of the subsystem, while the blue points indicate the predictions made by our machine learning method.