Optimization of tripartite quantum steering inequalities via machine learning.

We investigate the possibility of optimizing genuine tripartite quantum steering inequalities via machine learning. In particular, we consider two types of hybrid scenarios: one-sided device-independent scenario (where one party is considered to be untrusted) and two-sided device-independent scenario (where two parties are considered to be untrusted). In both scenarios, we apply a method of machine learning, known as artificial neuron networks, to optimize the quantum steering inequalities for the family of noisy Greenberger–Horne–Zeilinger states and noisy W states. The results show that the optimized steering classifiers can verify quantum steering well with only a small number of Pauli measurements, which can be easily realized in experiment. The method can be generalized to other multipartite or high-dimensional quantum systems. [ABSTRACT FROM AUTHOR]