Hybrid Multi-Stage Framework for Advanced Malware Detection: Enhancing Accuracy & Resilience

Malware detection is one of the parts of the modern cybersecurity system that requires precision and scalability against the ever-evolving threats. This paper will present a multi-staged hybrid framework for malware detection that combines static analysis, machine learning, and rule-based approaches. The files are processed through a sequence of stages: hash-based matching, machine learning classification, YARA rule evaluation, and decompiled code analysis. This further helps in accuracy and reduces the false prediction. The last decision will aggregate the outputs of all the stages using weighted scoring and achieving the final overall accuracy as 94.8% with a dataset of 8,932 samples. A layered approach ensures that a comprehensive analysis is provided along with a strong, scalable solution for malware detection. Our findings indicate that the multiplexing of various methods for detection may significantly boost the accuracy of detection along with minimizing the shortcomings inherent in each technique.