ID	原文	译文
58358	其次，利用降维处理将数据映射到三维空间，以便直观地可视化呈现和快速地进行数据分析;	Secondly，the datasets are mapped to a three-dimensional space through dimensionality reduction，whichis beneficial for intuitive visualization and rapid data analysis.
58359	最后，利用聚类分析和计算各源 IP 的可信度，检测出异常的源 IP.	Finally，clustering the source IPs and calculating the credibility of them to identify the abnormal ones.
58360	实验结果表明，所提算法不但能直观观察到多维数据集中的关联特性，而且能从全局和局部 2 个层面识别网络中异常的源 IP.	The experiment results show that this algorithm can not only observe the correlation characteristics of multi-dimensional datasets directly，but alsoidentify the abnormal source IPs in the global and local aspects.
58361	针对第 5 代移动通信系统( 5G) 网络切片映射过程中，在满足系统时延要求的情况下，使资源调度最优化的问题，提出了一种基于时延感知的 5G 网络切片节点和链路映射成本最小化算法.	To optimize the resource scheduling while meeting the system delay requirement，the articleproposes a delay-aware the fifth generation of mobile communications system ( 5G) network slicing nodeand link embedding algorithm in the process of 5G network slice embedding.
58362	该算法在网络功能虚拟化管理和编排器及各网络功能服务器处建立两级队列动态调度模型，感知系统中当前队列积压状态并进行动态调度，使系统队列积压始终维持在稳定的较小值，采用 Lyapunov 随机优化方法，实现对映射成本与系统时延的平衡控制.	The algorithm establishes atwo-level queue dynamic scheduling model at the network functions virtualization management and orchestration and network functions virtualization servers. It realizes a current queue backlog in the system andcarries on the dynamic scheduling，so that the system queue backlog is always maintained at the stablesmaller value. The algorithm achieves the balance control between the embedding cost and the system delay by Lyapunov stochastic optimization method.
58363	仿真结果表明，所提算法可在满足系统时延要求的同时，最优化资源调度，进而使得 5G 网络切片映射成本最小.	The simulation results show that the algorithm can optimize the resource scheduling while satisfying the system delay requirement，and minimize 5G networkslice embedding cost as well.
58364	针对文本信息隐藏嵌入容量低和语义连贯性差的问题，提出了一种基于神经网络图像描述的文本信息隐藏模型.	Aiming at the problem of low embedding capacity and poor semantic coherence of text steganography，a text steganographic scheme based on neural image caption is proposed.
58365	将卷积神经网络与长短期记忆网络相结合，把图像特征和生成语句进行关联.	An encode-decodestructure with a combination of long short term memory and convolution neural network is used to modelthe joint probability distributions between image features and the descriptive sentences.
58366	从收发双方能否共享图像及模型参数的不同应用前提出发，设计了多种概率采样方式，从而生成载密的图像描述文本.	Two methods withdifferent sampling process are designed from the perspectives of sharing and non-sharing models.
58367	实验结果表明，该算法具有较高的隐藏容量，载密描述句能较好地表达图像内容.	Experimental results show that the proposed model can achieve high embedding capacity and desirable text quality.