Abstract:
Multiple sophisticated malware attacks in the cyber world now present substantial danger to individual users as well as business operations and vital infrastructure systems. Due to quick-moving threats signature-based malware detection methods lose their effectiveness so organizations need machine learning techniques to ensure better accuracy and adaptability. The project establishes an AI-controlled malware detection solution based on several machine learning algorithms including Random Forest alongside XGBoost and Decision Tree and Logistic Regression together with K Nearest Neighbors (KNN) and Support Vector Machine (SVM). The dataset employed for model training and assessment was acquired from Kaggle after implementing numerous preprocessing and data balancing and feature selection strategies with SMOTE. The test results revealed that Random Forest delivered the best performance with a 99.21% accuracy rate during these evaluations. The evaluation of models to detect malware opposed to legitimate files was conducted using performance metrics such as precision, recall, F1-score alongside confusion matrices. The research revealed several drawbacks linked to its high accuracy rate including inaccurate results and technical difficulties related to model application together with data set dependency. Future research on these subjects will address malware monitoring in real time as well as deep learning and hybrid analysis and explainable AI to improve security and interpretation capabilities. The research demonstrates how modern cybersecurity relies heavily on AI technology together with machine learning for detecting malware threats through intelligent scalable solutions. Strategic integration of superior detection technologies between organizations and individuals helps reduce digital attack risks which leads to enhanced digital security
