Research on the Application of High-Dimensional Feature Fusion and Machine Learning Algorithms in Malicious Document Detection

With the rapid development of Internet technology, the dissemination of malicious documents has become increasingly rampant, and traditional detection methods for malicious documents have failed to meet the current complex and diverse security requirements. The high-dimensional features of malicious documents, including instruction-level, function-level, PE header information, and byte-level code, pose challenges for detection. This paper deeply investigates the application of high-dimensional feature fusion and machine learning algorithms in malicious document detection. Through feature extraction and fusion techniques, key features of malicious documents are extracted from multiple dimensions, and advanced machine learning algorithms such as Convolutional Neural Networks (CNN), Support Vector Machines (SVM), Random Forests (RF), and LightGBM are used to model and analyze the fused high-dimensional features. Experimental results show that the combination of high-dimensional feature fusion and machine learning algorithms significantly improves the accuracy and efficiency of malicious document detection, providing effective means to address increasingly complex network security threats.