Adaptive Cyber Attack Prediction Using Multi-Source Data

Cyber threats are evolving rapidly, making traditional detection systems inadequate for identifying sophisticated attack patterns. This paper presents an adaptive deep learning approach for cyber threat detection using multi-source textual data. The proposed system leverages Long Short-Term Memory (LSTM) networks to analyze and classify cyber threat intelligence into multiple categories, including phishing, malware, ransomware, and SQL injection. Advanced Natural Language Processing (NLP) techniques, such as lemmatization, stopword removal, and text normalization, enhance the model's ability to extract meaningful features from raw text. The system is trained on a diverse dataset comprising security alerts, phishing emails, and malware reports. Experimental results demonstrate that the model achieves an accuracy of 92-95%, outperforming traditional rule-based and machine learning-based detection methods. The findings indicate that deep learning-based cyber threat classification provides a scalable and effective solution for real-time cybersecurity defense. Keywords: Cyber Threat Detection, Deep Learning, Long Short-Term Memory (LSTM), Natural Language