ID	原文	译文
3063	系统仿真与分析表明，所提方法能够有效地抑制噪声的影响以及对多普勒引起的时间相关性进行提取。	System simulation and analysis show that the proposed method can effectively suppress the in-fluence of noise and extract the time correlation caused by Doppler.
3064	与代表性 CSI 压缩反馈方法和 CsiNet 方法相比，所提方法拥有更好的归一化均方误差（NMSE）和余弦相似度性能。	Compared with the traditional CSI compressionfeedback algorithm and CsiNet algorithm, the proposed method has better NMSE and cosine similarity performance.
3065	为了降低面向再生编码分布式存储系统的外包数据审计机制的安全实现开销，提出了一种正交化代数编码方法，以此构造一类基于线性同态认证的轻量型隐私保护审计方案。	To reduce the security implementation cost of the outsourcing data audit mechanism for the regenerat-ing-coding-based distributed storage systems, an orthogonal algebraic coding method was put forward to construct a lightweight privacy-preserving audit scheme based on linear homomorphic authentication.
3066	利用文件编码数据与私有密钥向量的正交化构造外包存储向量的同态认证标签，并提出利用密钥特定分量的正交基向量组的随机化掩码来完成审计响应消息的隐私保护，实现代数编码、隐私保护和安全审计的高效融合。	The homomorphic authentication tags were gen-erated with the orthogonalization between the file encoded data and the private secret key vector, and the privacy protection of the auditing response message was achieved by using the random masking that was constructed by randomizing the orthogonalbasis vectors of the specific sub-vector of the user's secret key. The work realized the effective integration of algebraic coding,privacy protection, and security auditing.
3067	理论分析表明，所提方案在再生编码分布式存储应用中可实现信息理论意义下的安全性。	Theoretical analysis shows that the proposed scheme can realize the infor-mation-theoretic security in the regenerating-coding-based storage applications.
3068	与现有同类工作相比，该方案计算复杂度低，通信开销小，具有更好的性能优势。	Compared with the existing works, the proposedscheme is of low computational complexity and communication overhead, and better performance advantages.
3069	针对说话人确认中无法准确捕获特征间非线性关系的问题，提出了一种基于深层信息散度最大化的目标函数表示方法。	To solve the problem that the nonlinear relationship between speaker representations cannot be accurately captured in speaker verification, an objective function based on depth information divergence maximization was proposed.
3070	该方法能通过计算特征所在分布之间相似度，来对特征间的非线性关系进行隐性表示，	It could implicitly represent the nonlinear relationship between speaker representations by calculating the similarity be-tween their distributions.
3071	并在最大化这种统计相关性的优化目标指导下，使深度神经网络向着同类数据更紧凑、异类数据更分散的方向优化，最终达到提升深层特征空间区分性的目标。	Under the supervision of the optimization goal of maximizing the statistical correlation, thedeep neural network was optimized towards the direction that the within-class data was more compact and the be-tween-class data were far away from each other, and finally the discrimination of deep speaker representation space couldbe effectively improved.
3072	实验结果表明，相对于其他深度学习方法，所提方法的相对等错误率（EER）最多降低了 15.80%，显著提升了系统性能。	Experimental results show that compared with other deep learning methods, the relative EER of the proposed method is reduced by 15.80% at most, which significantly improves the system performance.