Machine Learning Based Approach for Malware Classification

Document Type

Conference Proceeding

Source of Publication

2026 17th Student Research Conference on Applied Computing Src 2026

Publication Date

4-22-2026

Abstract

The increasing complexity of malware has reduced the effectiveness of traditional signature-based detection systems. While machine learning and deep learning techniques have emerged, many existing studies overlook the impact of class imbalance or lack a comprehensive comparison between different architectures under identical evaluation settings. These limitations affect the reliability of reported results for real-world malware detection. This paper presents a comparative study of Naïve Bayes, Support Vector Machine, feed forward Neural Network, and a one-dimensional Convolutional Neural Network (1D-CNN) for malware detection using the publicly available Malware Analysis Datasets: API Call Sequences. The models are evaluated under identical experimental settings, both with and without the Synthetic Minority Over-sampling Technique (SMOTE), to assess the impact of class imbalance on detection performance. Experimental results show that SMOTE reduces bias toward the majority class and improves performance for the minority class. Deep learning models achieve the highest performance after applying SMOTE. The feed forward neural network achieved an F1 score of 0.8180. The 1D-CNN outperforms all other models, achieving balanced performance with an F1 score of 0.8249, precision of 0.8184, and recall of 0.8317. These results show that imbalance-Aware training combined with deep learning models improves detection performance.

ISBN

[9798319510167]

Publisher

IEEE

Disciplines

Computer Sciences

Keywords

Behavioral Analysis, Class Imbalance, Cybersecurity, Deep Learning, Machine Learning, Malware Detection, SMOTE

Scopus ID

105043012832

Indexed in Scopus

yes

Open Access

no

Link to Full Text

Share

COinS