CYBER THREAT DETECTION IN CLOUD BASED ON ARTIFICIAL NEURAL NETWORK USING EVENT PROFILES

Abstract

Developing an automated method for detecting cyberthreats is one of the main issues facing 
cyber security. In this paper, we describe an artificial neural network-based cyberthreat detection 
method. The suggested solution improves cyber-threat identification by converting a large 
number of gathered security events into unique event profiles and utilizing a deep learning-based 
detection algorithm. In this study, we created an AI-SIEM system using a variety of artificial 
neural network techniques, such as CNN, LSTM, and FCNN, together with event profiling for 
data preparation. The system's primary objective is to distinguish between true positive and false 
positive alarms, enabling security analysts to act quickly. to online dangers. The authors of this 
work conducted all of the experiments using two real-world datasets and two benchmark 
datasets, NSLKDD and CICIDS2017. We ran trials utilizing the five traditional machinelearning techniques (SVM, k-NN, RF, NB, and DT) to assess the performance comparison with 
current methodologies. As a result, the study's experimental findings confirm that our suggested 
approaches can be used as learning-based models for network intrusion detection and 
demonstrate that, despite being used in real-world scenarios, their performance surpasses that of 
traditional machine learning techniques.