ID 原文 译文
44686 为了加强对 Android 恶意软件检测系统识别对抗样本的能力，提出了基于深度卷积生成对抗网络的Android 对抗样本生成框架。 In order to enhance the ability of Android malware detection system to identify adversarial samples, an Android adversarial sample generation framework was proposed based on deep convolution generation adversarial network.
44687 此框架模拟了恶意软件制作者的攻击行为，并在提出的 ASG 算法的帮助下实现对恶意软件的修改。 This framework simulates the attack behavior of malware makers and implements modifications to malware with the help of proposed ASG algorithm.
44688 此框架生成的恶意软件可以绕过检测系统的检测，并能够实际运行而不影响其原有恶意功能。 Malware generated by this framework can bypass the detection system and can actually run without affecting its original malicious functions.
44689 生成的对抗样本可用于重训练原始检测系统，提高系统应对对抗样本的能力。 The generated adversarial samples can be used to retrain the original detection system and improve the system's ability against adversarial samples.
44690 在 IDS 中，传统数据分组的捕获是从网卡复制到内核，再由内核复制到用户空间， In IDS, the traditional methods of packet captures are that packets are copied from the network card to the kernel and then they are copied to the user space, which results in frequent CPU interrupt response, redundant data replication, and context switching.
44691 这导致了 CPU 频繁地中断响应、冗余数据复制和上下文切换，没有充足时间来进行数据分组的进一步处理。 Then the system haven't sufficient time for further packet processing, which results in reducing performance.
44692 为了提升捕获包效率，采用多线程思想，通过 PF_RING ZC 技术实现零拷贝，把 PF_RING ZC 捕获数据分组的方法做成动态链接库，并集成到 Snort 中； In order to improve the efficiency of capturing packets, Zero-copy technology, based on the architecture of multi-thread which was implemented by PF_RING ZC method, and the method of packets capture is made into a dynamic link library and integrated into Snort.
44693 对捕获技术进行 IPv6 协议扩充，使 IDS 实现了支持 IPv6 检测的功能。 For implementing detection IPv6 packet in IDS, the expanded capture technology of IPv6 protocol was proposed.
44694 实验表明，相比 libpcap技术，PF_RING ZC 技术在高速和低速网络环境中有着更低的丢包率和 CPU 占用率。 The result of experiments shows that, compared with the method libpcap of packet capture, PF_RING ZC method has lower packet loss rate and lower CPU utilization rate in both high-speed and low-speed network environments.
44695 分析了目前 DNS 隧道检测各种方法，重点研究基于机器学习技术的 DNS 隧道分类方法，针对目前 DNS隧道检测只局限于特定隧道类别进行判别的不足，提出了采用多种分类算法进行组合分类决策的混合分类算法模型（CCAM）对 DNS 隧道进行检测与分类，CCAM 算法采用了支持向量机（SVM）、朴素贝叶斯（NB）、决策树（DT）等 3 种机器学习分类算法进行混合分类、组合训练与加权求优。 To propose a high precision detection method for DNS channel, different techniques for DNS tunneling detection were surveyed, the machine learning techniques based detection for DNS tunneling was researched, the existing problems of the current DNS tunneling detection based of specific tunneling classification was analyzed, combined classification algorithm model (CCAM) was introduced to classify DNS tunneling, CCAM uses SVM,NB and DT.