一种基于堆叠的深度可分离的卷积和注意力机制的恶意代码检测方法
Hong Huang1, Rui Du1, Zhaolian Wang1
1School of Computer Science and Engineering, Sichuan University of Science & Engineering, Yibin 644002, China.
Sensors (Basel, Switzerland)
|August 26, 2023
概括
这项研究介绍了CoAtNet,一种新的恶意软件检测方法,使用堆叠的深度可分离卷曲和自我注意力. 与现有模型相比,CoAtNet提高了恶意软件检测准确度和概括能力.
科学领域:
- 计算机科学 计算机科学
- 网络安全 网络安全
- 人工智能的人工智能
背景情况:
- 传统的恶意软件检测方法在较弱的模型概括和有限的容量适应方面扎.
- 现有的方法往往无法有效平衡模型的概括和适应.
- 需要先进的技术来提高恶意软件检测的稳定性.
研究的目的:
- 提出一种新的恶意软件检测方法,CoAtNet,将堆叠的深度可分离卷曲和自我注意力结合起来.
- 增强在恶意软件检测中的模型通用化和能力适应.
- 为了验证CoAtNet对已建立的恶意软件检测模型的有效性.
主要方法:
- 恶意代码被转换成灰度图像进行处理.
- 使用了一种采用堆叠深度可分离卷积和自我注意机制的检测模型.
- 使用Malimg和增强的Blended+数据集进行了比较实验.
主要成果:
- 与XceptionNet,EfficientNetB0,ResNet50,VGG16,DenseNet169和InceptionResNetV2.2.2相比,CoAtNet在准确性和概括性方面表现出了卓越的表现,这些表现都与XceptionNet,EfficientNetB0,ResNet50和InceptionResNetV2.2.2相比.
- 该模型有效地从恶意软件图像中提取必要的功能.
- 实验验证证证实了拟议方法的增强功能.
结论:
- CoAtNet 方法有效地解决了传统恶意软件检测的局限性.
- 堆叠的深度可分离卷积和自我注意力显著提高了检测准确性和稳定性.
- 这项研究为先进的恶意软件检测系统提供了一个有希望的解决方案.
相关概念视频
Detection of Black Holes
2.2K
Although black holes were theoretically postulated in the 1920s, they remained outside the domain of observational astronomy until the 1970s.
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
2.2K
Masking and Demasking Agents
2.5K
EDTA titrations may necessitate masking and demasking agents to temporarily protect a particular metal ion in a mixture from the EDTA reaction. These agents facilitate the sequential analysis of the metal ions by forming stable complexes with some—but not all—metal ions during certain steps.
There are many masking agents, such as cyanide, fluoride, triethanolamine, thiourea, and 2,3-bis(sulfanyl)propan-1-ol (formerly 2,3-dimercapto-1-propanol), with the masking agent chosen based on...
There are many masking agents, such as cyanide, fluoride, triethanolamine, thiourea, and 2,3-bis(sulfanyl)propan-1-ol (formerly 2,3-dimercapto-1-propanol), with the masking agent chosen based on...
2.5K


