Authors:
(1) SM Rakib Hassan, Computer Science and Engineering Department, Barak UniversityDhaka, Bangladesh ([email protected]);
(2) Aker Dakal, Computer Science and Engineering Department, Barak University, Dhaka, Bangladesh ([email protected]).
Links table
Summary and I. Introduction
the second. Literature review
Third. methodology
Fourth. Results and discussion
Fifth. Conclusion, future work, references
**a summary** In the age of the Internet and smart devices, the discovery of harmful programs has become very important for the security of the system. The authors of harmful programs are increasingly used jamming techniques to evade advanced security solutions, making it difficult to discover and eliminate threats. Excessive malware, which extends to hiding themselves, is a great risk on various platforms, including computers, mobile devices and Internet of Things. It fights traditional methods such as guidelines or signing against this type of malware, as they leave no clear effects on the system. In this research, we suggest the system of detecting simple and cost -effective harmful programs by analyzing memory emptying, using the diverse machine learning algorithms. The study focuses on the CIC-Mammem-2022 data collection, designed to simulate real world scenarios and evaluate the detection of memory-based malware. We assess the effectiveness of automated learning algorithms, such as decision -making, group methods, and nerve networks, in discovering excessive malware inside memory dumps. Our analysis extends multiple harmful software categories, providing insight into the strengths and restrictions of the algorithm. By providing a comprehensive assessment of automated learning algorithms to discover excessive malware through memory analysis, this paper contributes to continuous efforts to enhance cybersecurity and fortify digital ecosystems against advanced and advanced harmful threats. The source code is directed to open access to reproduction and future research endeavors. It can be accessed on https://bit.ly/malmemcode
I. Introduction
The rise of the Internet and smart devices has transformed many sectors, but also led to a sophisticated scene of threat, including the interconnected systems of advanced harmful programs. Excessive harmful programs, the skill in hiding themselves, is a major challenge to traditional cyber security methods. Traditional systems based on instructions or signature to identify such void threats, which requires a shift towards innovative and adaptive detection mechanisms.
This paper explores the discovery of excessive malware through a multi -sectarian classification, with the aim of bridging the gap between advanced threats and advanced methods of detection using machine learning. We analyze many algorithms, including decision trees, band roads, support vessels, and nerve networks, to detect their capabilities and restrictions in determining excessive malware.
Acknowledging the importance of class balance in data groups in the real world, especially in detecting harmful programs, we verify techniques such as sampling offers (edited by the nearest neighbor base, near the base, randomly under samples, and all KNN under samples) and generation Industrial data using the Adasyn method to process this challenge.
Our research simulates, based on the CIC-Mammem-2022 data collection, scenarios in the real world to detect memory-based malicious programs. By analyzing automatic learning algorithms accurately and data budget techniques, we contribute to the fortification of cybersecurity against advanced malicious threats.
In the following sections, we delve deeper into the collection of data, methodologies and results, we aim to provide valuable visions that can constitute the future of strategies for detecting malicious programs and cyber security amid the challenges offered by harmful programs and sudden clarity.
