ID 原文 译文
56508 最后,在真实数据集上进行大量实验,结果充分证明本文提出的模型在油气储集层识别任务上的有效性和鲁棒性. A large number of experimentson real industrial datasets are subsequently performed. The results fully prove the effectiveness and robustnessof the proposed model in oil and gas reservoir detection.
56509 机器学习方法对网络流量分类的前提是假设流量具有独立同分布性,而实际情况下流量特征不断发生变化,导致该方法在处理海量、不具备独立同分布的流量数据时开销较大,计算复杂度较高,精度较低. The prerequisite of traffic classification based on machine learning models is that traffic is independentand has identical distributions. However, traffic changes in the wild increase the memory cost of these modelsand reduces their accuracy.
56510 针对上述问题,本文提出一种新的分类模型. To tackle these problems, this work proposes a new classification model.
56511 该模型将PCA算法与改进的深度卷积神经网络分类模型(improved deep LeNet-5 convolutional neural networks, LCNN)相结合进行流量分类. The modelcombines a principal component analysis algorithm and an improved deep convolutional neural network.
56512 前者进行降维分析,发现影响检测精度的关键特征,后者采用自主特征学习方式提升分类精度. Theformer performs dimensionality reduction so that the key features affecting detection accuracy are found. Thelatter adopts the autonomous feature learning method to improve the classification accuracy.
56513 实验表明,本文方法的内存开销较之前方法降低了3.2%,检测精度提升了5%～8%. Experiments showthat the memory overhead is reduced by 3. 2% and that the detection accuracy is improved by 8% relative toother similar works.
56514 安全多方计算是国际密码学界研究的热点,百万富翁问题是安全多方计算最基础最重要的问题,是构造其他安全多方计算协议的基本模块. Secure multiparty computation is a focus of the international cryptographic community. The mil?lionaires’ problem is the most important problem in secure multiparty computation and is a building block forconstructing other secure multiparty computation protocols.
56515 这个问题已经有许多解决方案,但除了基于混淆电路的协议之外,目前基于公钥加密算法的解决方案几乎都是半诚实模型下的解决方案,抗恶意敌手的解决方案极少,仅有的个别解决方案效率很低,这制约着恶意模型下许多安全多方计算问题的解决. Several solutions are available to solve this problem,but except for protocols based on garbled circuits, the existing solutions based on public key cryptosystems areonly secure in semihonest models. No solution based on a public key cryptosystem is secure against maliciousadversaries. This state restricts the resolution of many secure multiparty computation problems in maliciousscenarios.
56516 抗恶意敌手的解决方案更符合安全多方计算的实际应用场景,研究抗恶意敌手的百万富翁问题解决方案,具有重要的理论与现实意义. A solution that is secure in malicious models is highly applicable in practical application scenariosand is generally appealing. Therefore, the study of the solution to the millionaires’ problem in a malicious modelis of great theoretical and practical significance.
56517 本文首先设计了一个半诚实模型下百万富翁问题的解决方案,进一步分析了恶意敌手可能的恶意行为,并用零知识证明和分割选择阻止或发现这些恶意行为,将半诚实模型下安全的计算协议改造成恶意模型下安全的计算协议,并用理想–实际范例证明了协议的安全性,分析了恶意敌手攻击成功的概率和方案的效率. In this work, we propose a multiparty computation protocolfor the millionaires’ problem that is secure in a semihonest model. The proposed protocol is simple and easilyunderstandable. We analyze the possible malicious behaviors in this protocol and use zero-knowledge proof andcut-and-choose techniques to resist possible malicious behaviors and thereby convert the protocol into one thatis secure in the malicious model. We prove that the proposed protocol is secure in the malicious model by usingthe well-accepted ideal-real paradigm.