ID 原文 译文
2893 哈希证明系统由 Cramer-Shoup 在 2002 年首次提出，到目前为止仍是密码工作者的研究热点之一。 Hash proof systems, which was first introduced by Cramer and Shoup in 2002, is still one of the hottest re-search topics in cryptography.
2894 进而，Wee 在 2010 年提出可提取哈希证明系统的概念，其可用来构造基于查找性困难假设的公钥加密方案。 And then Wee proposed the concept of extractable hash proof system in 2010 and it is a con-cept extension on the hash proof system and as a paradigm of constructing PKE from search problems.
2895 本文在可提取哈希证明系统之上，通过重新定义系统参数的意义，扩大了可提取哈希证明系统的密码学应用范围。 On the basis of theextractable hash proof system, this paper expands the application scope of the extractable hash proof system by redefining themeaning of system parameters.
2896 我们利用可提取哈希证明系统的框架构造了一个基本的基于 Diffie-Hellman 关系的 All-But-One 可提取哈希证明系统。 We construct a basic All-But-One extractable hash proof system based on Diffie-Hellman re-lations by using the framework of extractable hash proof system.
2897 在此基础上细粒度了辅助输入，引入权重计算，给出了一个基于标签和可变策略的 CCA 加密方案，并进行了详细的安全性证明。 Based on this, fine-grained auxiliary input and weighting calculation are introduced. A new variable-policy CCA encryption scheme based on tag is proposed, and the security proof isalso given in details.
2898 特别的，该方案比可提取具有更丰富的抽象表达，即是 All-But-N 的，也即在提取模式中由标签决定的分支数量可以有 n个。 In particular, this scheme is a richer abstraction of extractable hash proof system that it is All-But-N, which means that the number of branches determined by the tag in the extraction mode could be n.
2899 同时，该方案是基于困难性可搜索问题，本质上是基于计算性的 Diffie-Hellman 问题。 At the same time, the scheme is based on the difficulty of the search problem and is essentially based on the computational Diffie-Hellman prob-lem.
2900 行人再识别问题中，由于视角、光照和行人姿态等因素的变化，导致难以提取有效的行人特征，降低识别精度。 It is challenging to learn efficient features in person re-identification task due to complex variations of viewpoints, illumination, pose etc.
2901 而深度神经网络在训练样本较少的情况下较难训练，易出现过拟合现象。 In addition, deep neural network still suffers from overfitting with a small training set.
2902 针对上述问题，本文提出一种多信息流动卷积神经网络(Multi-information Flow Convolutional Neural Network，MiF-CNN)模型，模型中包含一个特殊的卷积结构， To solve these problems, a Multi-information Flow Convolutional Neural Network(MiF-CNN)is designed for person re-identifi-cation which contains a specific convolutional architecture.