Cyber Threat Intelligence System Using Cnn-lstm Deep Learning Model

• Author(s):

  Sai Santhosh A | Pravinkumar P | Rishikesh GC | Raghavan B

• Keywords:

  Cyber Threat Intelligence, Deep Learning, CNN-LSTM Model, Network Security, Intrusion Detection, Malware Detection, DDoS Attack, Feature Extraction, Sequential Data Analysis, Cybersecurity, Threat Classification, Artificial Intelligence, Anomaly Detection

• Abstract:

  Cyber Threat IntelligenceThe Rapid Evolution Of Cyber Threats, Including Malware, Phishing Attacks, Distributed Denial Of Service (DDoS), And Advanced Persistent Threats, Has Made Traditional Security Mechanisms Less Effective. Conventional Rule-based And Signature-based Detection Systems Fail To Identify Unknown Or Zero-day Attacks, As They Rely Heavily On Predefined Patterns. This Limitation Highlights The Need For Intelligent And Adaptive Cyber Threat Intelligence (CTI) Systems Capable Of Analyzing Large-scale And Complex Data In Real Time. This Project Proposes A Cyber Threat Intelligence System Using A Hybrid CNN-LSTM Deep Learning Model To Enhance Threat Detection And Classification. The System Combines The Strengths Of Convolutional Neural Networks (CNN) And Long Short-Term Memory (LSTM) Networks. CNN Is Utilized For Efficient Feature Extraction From High-dimensional Network Traffic Data, Identifying Important Spatial Patterns Such As Anomalies And Malicious Signatures. LSTM, A Type Of Recurrent Neural Network (RNN), Is Employed To Capture Temporal Dependencies And Sequential Patterns In The Data, Which Are Essential For Detecting Evolving And Time-based Cyber Attacks. The System Processes Large Cyber Security Datasets, Such As Network Traffic Logs, By Performing Data Preprocessing Steps Including Normalization, Noise Removal, And Feature Transformation. The Processed Data Is Then Fed Into The CNN-LSTM Model For Training And Prediction. The Model Classifies Network Activities Into Multiple Categories Such As Normal Traffic, Intrusion Attempts, Malware, And DDoS Attacks. Experimental Results Demonstrate That The Proposed Hybrid Model Achieves Higher Accuracy, Better Precision, And Reduced False Positive Rates Compared To Traditional Machine Learning Models Like Support Vector Machines (SVM) And Decision Trees. Additionally, The System Is Capable Of Handling Large-scale Data Efficiently, Making It Suitable For Real-time Threat Detection Environments.

---

Other Details

• Paper id:

  IJSARTV12I4105063

• Published in:

  Volume: 12 Issue: 4 April 2026

• Publication Date:

  2026-04-18

Download Article