ID 原文 译文
18045 生成的样本在有效对抗基于机器学习的恶意代码检测模型的同时，确保了恶意代码样本的可执行和恶意行为的一致性，有效提升了生成对抗样本的真实性和模型对抗评估的准确性。 The generated samples combat effectively the malicious code detection model based onmachine learning, while ensuring the consistency of the executable and malicious behavior of malicious codesamples, and improving effectively the authenticity of the generated adversarial samples and the accuracy ofthe model adversarial evaluation are presented.
18046 实验表明，该文提出的对抗样本生成方法使MalConv恶意代码检测模型的检测准确率下降了14.65%；并可直接对VirusTotal中4款基于机器学习的恶意代码检测商用引擎形成有效的干扰， The experiments show that the proposed method of generatingadversarial samples reduces the detection accuracy of the MalConv malicious code detection model by 14.65%, and can directly interfere with four commercial machine-based malicious code detection engines in VirusTotal.
18047 其中，Cylance的检测准确率只有53.55%。 Among them, the accuracy rate of Cylance detection is only 53.55%.
18048 安卓系统越来越广泛地被应用于各种类型的智能设备，比如智能手机、智能手表、智能电视、智能汽车。 Android system is now increasingly used in different kinds of smart devices, such as smart phones,smart watches, smart TVs and smart cars.
18049 与此同时，针对这些平台应用软件的逆向攻击也日益增多，这不仅极大地侵犯了软件开发者的合法权益，也给终端用户带来了潜在的安全风险。 Unfortunately, reverse attacks against Android applications are also emerging, which not only violates the intellectual right of application developers, but also brings security risks to end users.
18050 然而，现有的安卓软件保护方法比如命名混淆、动态加载、代码隐藏等虽然可在一定程度上增加安卓软件的逆向难度，但是原理相对简单容易被绕过。 Existing Android application protection methods such as naming obfuscation, dynamic loading,and code hiding can protect Java code and native (C/C++) code, but are relatively simple and easy to be bypassed.
18051 一种更为有效的方法是基于指令虚拟化的加固方法，但已有的指令虚拟化方法只针对特定架构(x86架构)，无法兼容运行于多种架构的安卓设备。 A more promising method is to use instruction virtualization, but previous binary-based methodstarget specific architecture (x86), and cannot be applied to protect Android devices with different architectures.
18052 该文针对安卓应用软件中的本地代码提出了一种架构无关的指令虚拟化技术，设计并实现了基于虚拟机打包保护(VMPP)的加固系统。 An architecture-independent instruction virtualization method is proposed, a prototype named Virtual MachinePacking Protection (VMPP) to protect Android native code is designed and implemented.
18053 该系统包含一套基于寄存器架构的定长虚拟指令集、支持该虚拟指令集的解释器以及可以与现有开发环境集成的工具链。 VMPP includes aregister-based fix-length instruction set, an interpreter to execute virtualized instructions, and a set of tool-chains for developers to use to protect their code.
18054 在大量C/C++代码以及真实安卓软件上的测试表明，VMPP在引入较低的运行时开销下，能够显著提升安卓本地代码的防逆向能力，并且可被用于保护不同架构上的安卓本地代码。 VMPP is tested on a large number of C/C++ code and real-world Android applications. The results show that VMPP can effectively protect the security of Android nativecode for different architectures with low overhead.