Multimodal machine learning approach for detecting spyware and ransomware
Pidugu Trisandhya, Srishti Vashishtha, Udit Hasija, Himanshu Dadhwal, Reeba Qureshi · 2024
This study acquaints one with an encyclopaedic approach of the machine learning field which is distinctively designed for the identification of the risk imported by the two trenchant and widespread malwares, spyware and ransomware. With spyware and ransomware posing substantial threats to both individuals and businesses worldwide, recent reports underscore the urgency to solve these problems on an urgent basis. Check Point Research's findings reveal a 55.5% surge in attempted ransomware attacks compared to the previous year, locking the target over 1 in 10 organizations globally. Additionally, the Positive Technologies report for Q3 2023 highlights a 65% increase in spyware attacks on individuals compared to the prior quarter. This underscores the critical need for robust detection mechanisms. Our study aims to address this challenge by exploring multiple innovative approaches to malware detection. In our research, we found that Support Vector Machine (SVM) and XGBoost demonstrate the best performance, with accuracies of 99.7% and 99.6% respectively. We seek to further improve cybersecurity practices in the digital age by showcasing the scalability and efficacy of our suggested strategy in practical settings through experimental evaluation and analysis.