In this paper we initiate the study of the computational complexity of learning linear temporal logic (LTL) formulas from examples. We construct approximation algorithms for fragments of LTL and prove hardness results; in particular we obtain tight bounds for approximation of the fragment containing only the next operator and conjunctions, and prove NP-completeness results for many fragments.
翻译:在本文中,我们开始研究从实例中学习线性时间逻辑(LTL)公式的计算复杂性。我们为LTL的碎片构建近似算法,并证明其硬性效果;特别是,我们获得了仅包含下一个操作员和连接的碎片近似近距离的严格限制,并证明了许多碎片的NP完整性结果。