ID	原文	译文
2963	为了应对手工视觉特征与哈希编码过程不能最佳地兼容以及现有哈希方法无法区分图像语义信息的问题，提出一种基于深度卷积神经网络学习二进制哈希编码的方法。	To address the problems that hand-engineering visual features can't be optimally compatible with the Hashcoding process and existing Hash methods can't differentiate images semantics information, a learning method of binaryHashing based on deep convolutional neural networks is proposed.
2964	该方法基本思想是在深度残差网络中增加一个哈希层，同时学习图像特征和哈希函数;	The basic idea is to add a Hash layer into the deep residu-al network and to learn simultaneously image features and Hash functions.
2965	以此同时提出一种更加紧凑的分级哈希结构，用来提取更加接近图像语义的特征。	Meanwhile, we propose a more compact hierarchi-cal Hashing structure to extract features closer to semantics information of images.
2966	经 MNIST、CIFAR-10、NUS-WIDE 数据集的实验，结果表明该方法优于现有的哈希方法。	Experimental results of MNIST, CIFAR-10 and NUS-WIDE datasets show that the method is superior to existing Hashing methods.
2967	该方法不仅统一了特征学习和哈希编码的过程，同时深层残差网络也能得到更接近图像语义的特征，进而提高了检索准确度。	This method not only unifies theprocess of feature learning and Hash coding, and at the same time, the deep residual network is able to get features closer toimage semantics. Thus the retrieval accuracy is improved.
2968	针对现有可搜索加密领域所遇到的加密密钥维度高、更新不灵活和搜索速度慢等问题，我们提出了一种新型类别分组索引方法———CGIM。	In order to solve the problems such as high dimension of encrypted key, low degree of update flexibilityand low search speed in the field of encrypted search, we propose a novel classificatory group index method-CGIM.
2969	新方法将数据分类后，按类提取关键词建立分组索引，并采用分组加密方式实现以若干低维加密密钥代替高维加密密钥以缩短索引和查询请求的加密时间。	The method extracts category keywords from classified data to create group index, and uses group encryption method to transforma high-dimensional secret key into several low-dimensional keys to reduce the encryption time of indexes and query re-quests.
2970	此外，分组索引方法的每个组向量对应不同的类别，这样不仅可以实现分类更新以改善更新文档的灵活性，而且能够在检索过程中生成针对性陷门，从而进一步提高搜索的速度和效率。	In addition, each group in the index is corresponding to different category. Thus, the method can not only achieve classification update to improve the flexibility of document update, but also can generate a targeted trapdoor in the retrieval process to improve the search speed and efficiency further.
2971	理论和实验分析表明，该方法是可行且有效的。	Through security analysis and performance test, we prove that themethod is feasible and effective.
2972	为了抵消无人机“蜂群”所具有的非对称作战优势，从反控制其协同飞行的角度出发，将“蜂群”描述为具有涌现性特征的复杂系统，剖析无人机“蜂群”蜂拥涌现行为的产生机理，首次建立基于 f-散度的“蜂群”涌现性度量模型。	In order to neutralize the advantage of asymmetric operations owned by UAV swarms, UAV swarms wereinvestigated from the perspective of negativing the coordinated flight control for them. Firstly, UAV swarms are described as a complex system with the emergence. Secondly, based the mechanism for flocking behaviors generated from UAV swarms, f-divergence based quantitative model was established to measure the emergence of them.