ID 原文 译文
17555 再找出该矩阵在二元域中的零空间矩阵作为该码长下的监督矩阵，用不同长度码长的监督矩阵与待检测的码字迭代相乘，根据乘积结果中“1”的比例来判断码字的码长、信息位个数和位置分布。 Then, the null space matrix of this matrix in the binary field is found out as the supervision matrix under the code length.The code word is iteratively multipied by the supervision matrix of different code lengths, according to the proportion of "1" in the product result, the code length, number and position distribution of information bits of the code word are determined.
17556 仿真结果表明，针对200组码长64，信息位个数30的极化码，在最大误比特率不超过0.06时，识别率能保持在80%以上。 The simulation results show that for the 200 groups of polar code with 64-code-length and 30-information-bits, the recognition rate can be kept above 80% when the maximum bit error rate isless than 0.06.
17557 面对形态万千、变化复杂的海量极光数据，对其进行分类与检索为进一步研究地球磁场物理机制和空间信息具有重要意义。 It is of great significance to classify and retrieve the vast amount of aurora data with various forms and complex changes for the further study of the physical mechanism of the geomagnetic field and spatial information.
17558 该文基于卷积神经网络(CNN)对图像特征提取方面的良好表现，以及哈希编码可以满足大规模图像检索对检索时间的要求，提出一种端到端的深度哈希算法用于极光图像分类与检索。 In this paper, an end-to-end deep hashing algorithm for aurora image classification and retrieval isproposed based on the good performance of CNN in image feature extraction and the fact that hash coding canmeet the retrieval time requirment of large-scale image retrieval.
17559 首先在CNN中嵌入空间金字塔池化(SPP)和幂均值变换(PMT)来提取图像中多种尺度的区域信息； Firstly, Spatial Pyramidal Pooling(SPP) andPower Mean Transformtion(PMT) are embedded in Convolutional Neural Network (CNN) to extract multi-scaleregion information in the image.
17560 其次在全连接层之间加入哈希层，将全连接层最能表现图像的高维语义信息映射为紧凑的二值哈希码，并在低维空间使用汉明距离对图像对之间的相似性进行度量； Secondly, a Hash layer is added between the fully connected layer to MeanAverage Precision(MAP) the high-dimensional semantic information that can best represent the image into acompact binary Hash code, and the hamming distance is used to measure the similarity between the imagepairs in the low-dimensional space.
17561 最后引入多任务学习机制，充分利用图像标签信息和图像对之间的相似度信息来设计损失函数，联合分类层和哈希层的损失作为优化目标，使哈希码之间可以保持更好的语义相似性，有效提升了检索性能。 Finally, a multi-task learning mechanism is introduced to design the lossfuction by making full use of similarity informtion between the image label information and the image pairs.The loss of classification layer and Hash layer are combined as the optimization objective, so that a bettersemantic similarity between Hash code can be maintained, and the retrieval performance can be effectivelyimproved.
17562 在极光数据集和 CIFAR-10 数据集上的实验结果表明，所提出方法检索性能优于其他现有检索方法，同时能够有效用于极光图像分类。 The results show that the proposed method outperforms the state-of-art retrieval algorithms onaurora dataset and CIFAR-10 datasets, and it can also be used in aurora image classification effectively.
17563 针对当前关于服务功能链(SFC)的部署问题都未考虑到虚拟网络功能(VNF)的失效重要度，该文提出了基于深度强化学习的SFC可靠部署算法。 In view of the current deployment of the Service Function Chain (SFC), the failure importance of theVirtual Network Function (VNF) is not considered,an SFC reliable deployment algorithm based on deepreinforcement learning is proposed.
17564 首先建立VNF和虚拟链路可靠映射模型，为重要的VNF设置高可靠性需求，并通过链路部署长度限制尽可能保证虚拟链路可靠性需求。 Firstly, a reliable mapping model of VNF and virtual links is establised,high reliability requirements is set for important VNFs, and the reliability requirements of virtual links isensured as much as possible through link deployment length restrictions.