ID 原文 译文
50007 该方法基于软件体系结构代数建模思想，通过分析构件间交互的特点，使用代数范式形式抽象软件的基本结构风格。 The method based on algebraic modeling software architecture, through analyzing the characteristics of the interaction between components, using the basic structure of the form of abstract algebra paradigm software style.
50008 明确了范式向系统状态空间的映射关系，由此建立可靠性参数计算准则，并实现了系统可靠性评估的完整流程。 The mapping relationship of paradigm to the system state space, thus establishing the reliability parameter calculation rules, and implements the system reliability assessment of the integrity of the process.
50009 因为代数语言的高度形式化特征，流程具有结构嵌套处理以及自动完成计算的显著优点。 Because the height of the algebraic language formal features, process with a nested structure, processing and the obvious advantages of automatic calculation.
50010 最后通过对一个实际软件系统的可靠性分析，验证了代数方法的适用性与有效性。 Finally through the analysis of the reliability of a practical software system, to verify the applicability and effectiveness of algebra method.
50011 通过大脑对外界环境感知的神经结构与认知功能的相关研究，构建仿脑的媒体神经认知计算(multimedia neural cognitive computing，MNCC)模型。 Through the brain to the outside environment perception neural structures and cognitive function of related research, build media in imitation of a brain neural cognitive computing (multimedia neural cognitive computing, MNCC) model.
50012 该模型模拟了感官的信息感知、新皮层功能柱的认知功能、丘脑的注意控制结构、海马体的记忆存储和边缘系统的情绪控制环路等大脑基本的神经结构和认知功能。 The model simulates the function of sensory information perception, the neocortex column of cognitive function, the attention of the thalamus control structure, the hippocampus, memory storage and the mood of the limbic system control loop of the brain such as basic neural structures and cognitive function.
50013 在此基础上，构建基于MNCC的高分辨率遥感图像场景分类算法。 On this basis, build the high resolution remote sensing image scene classification algorithm based on MNCC.
50014 首先，图像经仿射变换后切分为若干图块，通过深度神经网络提取图块的稀疏激活特征，采用概率主题模型获取图块初始场景类别，并利用图块分类错误信息反馈控制场景显著区特征的提取; First, the image segmentation after affine transformation for some pieces, through deep sparse activation characteristics of neural network to extract the map, using probability model theme get figure of initial scene category, and use map block classification error feedback control scene area feature extraction;
50015 其次，根据图块的上下文获取场景语义的时空特征，并在此基础上进行图块分类和场景预分类; Secondly, according to the figure of the context for scenario semantic characteristics of space and time, and on this basis to classification of figure block classification and scene;
50016 最后，用场景预分类误差构造奖惩函数，控制和选择深度神经网络中场景区分度较大的稀疏激活特征， Finally, using scene recognition error structure is rewards and punishment function, control and choice for larger scenario in the neural network to distinguish the depth of sparse activation characteristics,