Malware API Calls Detection Using Hybrid Logistic Regression and RNN Model

Behavioral malware analysis is a powerful technique used against zero-day and obfuscated malware. Additionally referred to as dynamic analysis, this approach employs various methods achieve enhanced detection. One such method involves using machine learning deep algorithms learn from the behavior of However, task weight initialization in neural networks remains an active area research. In paper, we present novel hybrid model that utilizes both detect across categories. The proposed achieves by recognizing malicious functions performed malware, which can be inferred its API call sequences. Failure these instances result severe cyberattacks, pose significant threat confidentiality, privacy, availability systems. We rely on secondary dataset containing sequences, apply logistic regression obtain initial serves input network. By utilizing approach, our research aims address challenges associated with traditional techniques improve accuracy efficiency detection based calls. integration allows capitalize strengths each potentially leading more robust versatile solution Moreover, contributes ongoing efforts field networks, offering perspective their impact performance context behavioral analysis. Experimental results balanced showed 83% 0.44 loss, outperformed baseline terms minimum loss. imbalanced dataset’s was 98%, loss 0.10, exceeded state-of-the-art model’s accuracy. This demonstrates how well suggested handle classification.