We apply the artificial neural network to quantify two-qubit steerability based on the steerable weight, which can be computed through semidefinite programming. Due to the fact that the optimal measurement strategy is unknown, it is still very difficult and time-consuming to efficiently obtain the steerability for an arbitrary quantum state. In this work, we show the method via machine learning technique which provides an effective way to quantify steerability. Furthermore, the generalization ability of the trained model is also demonstrated by applying to the Werner state and that in dephasing noise channel. Our findings provide an new way to obtain steerability efficiently and accurately, revealing effective application of the machine learning method on exploring quantum steering.

中文翻译：

---

量化量子可操纵性的机器学习

我们应用人工神经网络基于可操纵权重量化两个量子位的可操纵性，该权重可通过半定编程来计算。由于最佳测量策略是未知的事实，有效地获得任意量子态的可操纵性仍然非常困难且耗时。在这项工作中，我们通过机器学习技术展示了该方法，该方法提供了量化可操纵性的有效方法。此外，通过应用到Werner状态和移相噪声通道，也证明了训练模型的泛化能力。我们的发现提供了一种有效而准确地获得转向性的新方法，揭示了机器学习方法在探索量子转向中的有效应用。