- Conference Article
1
- 10.1109/iscs69371.2025.11386404
Adaptive Cybersecurity Frameworks: Leveraging Artificial Intelligence for Proactive Threat Detection and Forensic Readiness
- Nov 14, 2025
- Mohit Garg + 5 more +5
Cyberattacks have been becoming increasingly sophisticated in speed and complexity, while Cybersecurity is being pushed by cyberattacks to provide the appropriate level of defensive preparedness. This research has developed an Adaptive Cyber Security Framework that uses artificial intelligence not only to identify potential cyber threats but to be prepared for the forensic analysis of those threats that are able to evade detection. The primary framework was based upon Long Short-Term Memory (LSTM) neural network architecture with attention mechanism, providing the model’s ability to provide strong predictive capabilities as well as a layer of interpretability which could be trusted by human security analysts. The simulation testing demonstrated the model was able to achieve a solid balance between precision and recall; demonstrated low latency performance even at high load levels; and remained stable and robust even when there were high levels of noise present within the test data. In addition, the model provided a satisfactory logging of anomalous events; demonstrating its readiness for forensic analysis. While the model was trained using synthetic data, the results indicate the development of a practical methodology for developing scalable, adaptable, and memory-based AI driven Cyber Defense systems rather than reactive ones.
Read more