Intrusion Detection for Predicting Security Attacks Using Hybrid LSTM-GRU Classifiers

Aggressive attacks, computer viruses, and malware routinely affect computer networks. Finding intrusions is one of the most crucial aspects of safeguarding the network and is a proactive defense technique. A novel method has to be provided for locating difficulties in cybersecurity and intrusions in cyber-physical system. This work proposed the hybrid Long Short Term Memory –Gated Recurrent Unit (LSTM-GRU) classifier for efficient intrusion detection system, which minimize modeling's computational cost and choose the best features. Both LSTM and GRU units are integrated into a single network in a hybrid LSTM-GRU design. The architectures of GRU and LSTM are capable of handling sequences of varying lengths. Because of this, hybrid LSTM-GRU models are often used for applications like time series analysis and natural language processing where the length of the input sequences fluctuates. This work is executed in Python software, which obtained a high detection training and validating data accuracy of 96.23% and 95.39% respectively.