基于预聚类主动半监督的作战体系效能评估

针对作战仿真实验中体系效能通常依靠专家评估、评估代价较大的问题，提出一种基于预聚类主动半监督学习的作战体系效能评估方法。明确了使用该方法进行作战体系效能评估的基本流程，以及自顶向下的评估模式和二值化的评估标准。重点构建了预聚类主动半监督学习算法，首先，结合作战仿真实验数据的特点，对未评估样本进行预聚类，选择最有价值的样本供专家标注；然后，使用已标注的样本训练主动学习算法和半监督学习算法的公用学习器；最后，利用主动学习算法挑选价值较高的样本交由专家评估，并利用新样本对学习器进行不断更新。作战仿真实验数据表明，该方法在达到预期评估准确度的同时降低了评估代价，能有效应用于大规模作战仿真实验的体系效能评估。

Evaluation of operational system of systems effectiveness based on pre-clustering active semi-supervised learning

Aiming at the problem that the measures of effectiveness (MOE) of system of systems in operational simulation experiment usually depends on expert evaluation with relatively high costs, this paper proposes an evaluation method of MOE based on pre-clustering active semi-supervised learning. This paper presents the basic process of using this method to evaluate the MOE of operational system of systems, the top-down evaluation model, and binarized evaluation criteria of this evaluation method. This paper focuses on constructing the pre-clustering active semi-supervised learning algorithm. Firstly, according to the characteristics of combat simulation experiment data, this method pre-clusters the unevaluated samples, and selects the most valuable samples for experts to mark. Secondly, it uses the evaluated samples to train the common learners of the active learning algorithm and the semi-supervised learning algorithm. Finally, it uses the active learning algorithm to select the high value samples for experts to evaluate, and it uses the new samples to continuously update the learners. Simulation results show that the method can reduce the evaluation cost while achieving the expected evaluation accuracy, and can be effectively applied to MOE evaluation of large-scale combat simulation experiments.