The systematism of weapon combat is the typical characteristic of a modern battlefield. The process of combat is complex and the demand description of weapon system of systems （SOS） is difficult. Granular analyzing ...The systematism of weapon combat is the typical characteristic of a modern battlefield. The process of combat is complex and the demand description of weapon system of systems （SOS） is difficult. Granular analyzing is an important method for solving the complex problem in the world. Granular thinking is introduced into the demand description of weapon SoS. Granular computing and granular combination based on a relation of compatibility is proposed. Based on the level of degree and degree of detail, the granular resolution of weapon SoS is defined and an example is illustrated at the end.展开更多
分类学习效果与有限训练样本的分布情况密切相关。支持向量数据描述(Support vector data description,SVDD)作为单一边界求解模型,不能良好刻画数据实际分布特征,从而导致部分目标对象落在超球以外。为了提高其分类能力,本文提出一种...分类学习效果与有限训练样本的分布情况密切相关。支持向量数据描述(Support vector data description,SVDD)作为单一边界求解模型,不能良好刻画数据实际分布特征,从而导致部分目标对象落在超球以外。为了提高其分类能力,本文提出一种基于粒计算的支持向量数据描述(Granular computing-driven SVDD,GrC-SVDD)分类方法,构造多粒度层次的属性集合以及相应的多粒度超球。首先通过邻域自信息对当前粒度层的属性集合重要度进行计算,然后选择最佳属性集合对上一粒度层未达到纯度阈值的超球再训练,直到所有超球满足条件或者属性耗尽。实验部分讨论了算法参数对分类性能的影响,并通过学习获得超参数。结果表明,与SVDD及流行的分类算法相比,本文方法具有较好的分类性能。展开更多
文摘The systematism of weapon combat is the typical characteristic of a modern battlefield. The process of combat is complex and the demand description of weapon system of systems （SOS） is difficult. Granular analyzing is an important method for solving the complex problem in the world. Granular thinking is introduced into the demand description of weapon SoS. Granular computing and granular combination based on a relation of compatibility is proposed. Based on the level of degree and degree of detail, the granular resolution of weapon SoS is defined and an example is illustrated at the end.
