ID 原文 译文
22115 以完整的 CT 肺部图像作为输入，池化层对输入图像进行降维处理； The input of the model is the complete CT lung image. The pooling layer reduces the dimension of input.
22116 批规范化解决随着网络深度的增加性能降低的问题； Batch normalization works out the poor performance with the increase of network depth.
22117 引入残差学习，学习模型中每一层的残差，最后输出去噪图像。 The residuals of each layer are learned with residual learning. Finally, the denoised image is produced.
22118 与经典去噪算法实验结果对比，所提方法在解决去噪方面达到了很好的滤波效果，同时也较好地保留了肺部图像的细节信息，大大优于传统的去噪算法。 Compared with classical methods, the proposed method achieves good filtering effect in solving the denoising method, and also retaining the details of lung image information, which is much better than the traditional filtering algorithm.
22119 移动 Ad hoc 网络(MANET)易遭受各种安全威胁，入侵检测是其安全运行的有效保障，已有方法主要关注特征选择以及特征权重，而忽略特征间潜在关联性，针对此问题该文提出基于图论的 MANET 入侵检测方法。 Mobile Ad hoc NETwork (MANET) is vulnerable to various security threats, and intrusion detection is an effective guarantee for its safe operation. However, existing methods mainly focus on feature selection and feature weighting, and ignore the potential association among features. To solve this problem, an intrusion detection method for MANET based on graph theory is proposed.
22120 首先通过对典型攻击行为分析，合理选择 9 种特征作为节点，依据欧式距离确定节点间的边以构建结构图。 First of all, nine features are selected as nodes based on the analysis of typical attack behavior, and the edges among nodes are determined according to Euclidean distance so as to build the structure diagram.
22121 其次发掘节点(即特征)间关联性，综合考虑节点邻居规模属性和节点邻居之间的紧密程度属性，利用图论所对应的统计特性度分布和聚集系数具体实现两属性。 Secondly, the scale attributes of neighborhood nodes and the degree of closeness attributes among nodes are considered to explore (i.e. feature) the correlation among nodes, then the statistical properties degree distribution and clustering coefficient of graph theory are used to realize the above two attributes.
22122 最后对比实验结果证明此方法与传统方法相比平均检测率和误检率分别提高10.15%、降低 1.8%。 Finally, contrasting experimental results show that compared with the traditional methods, the average detection rate and false detection rate of new method are improved by 10.15% and reduced by 1.8% respectively.
22123 针对稀疏阵列下 2 维波达方向(DOA)估计问题，该文提出一种基于稀疏采样阵列优化的加速逼近梯度(APG)算法与多重信号分类(MUSIC)算法相结合的 2D-DOA 估计方法。 A novel Two Dimension Direction Of Arrive (2D-DOA) estimation method based on sparse sampling array optimization is proposed, which is combined with Accelerated Proximal Gradient (APG) and MUltiple SIgnal Classification (MUSIC).
22124 首先，建立稀疏阵列下的 2D-DOA 估计信号模型，并证明其具备低秩特征，满足零空间性质(NSP)。 First, a 2D-DOA estimation signal model for sparse array is established, and its low rank feature and Null Space Property (NSP) are analyzed.