International Research Journal of Education and Technology

Computers, networks, and other resources are increasingly being harmed by malicious software, or malware. They are frequently disseminated via portable electronics and networks. Malware has a serious impact because to the sharp rise in internet usage. Malware developers keep coming up with new ideas in spite of improvements in detection systems. Internet users are very concerned about malware, a problem that is becoming worse. To evade identification by conventional detection models, polymorphic malware, a more versatile kind of malicious software, continuously modifies its signature characteristics. Malicious threats were identified using machine learning techniques; the optimal approach was indicated by high detection ratios. False positives and false negatives were measured by the confusion matrix, which offered more details on system performance. On a small FPR dataset, malware detection was successfully accomplished by the DT, CNN, and SVM algorithms. As the internet has grown, malicious software that steals data or eavesdrops has become more prevalent. Malware is defined by Kaspersky Labs as executable programs that are identified by machine learning methods. This study looks at a number of malware detection techniques, including as signature-based, heuristic, and behavior-based approaches. Conventional antivirus programs use signature-based detection, which is ineffective at spotting new or evolving threats. Heuristic analysis uses algorithms to provide security, whereas behavior-based detection monitors system behavior to find new threats. Detection capabilities are enhanced when machine learning and artificial intelligence are combined. Strong cybersecurity plans in a complicated digital environment require an understanding of these strategies.