In the ever -changing world of cyber security, an intrusion system is necessary to protect the network. However, Traditional signature -based infiltration detection systems often remember new or rapidly changing dangers when they depend on the pattern of the famous attack. This article suggests a hybrid deep learning model connecting an LSTM network and a region-based conversion nerve network to overcome these difficulties. While the LSTM network detects temporary dependence on traffic behavior, the component network removes spatial properties from traffic data.
The proposed approach detects successful complex infiltration patterns that change with timely and spatial convenience. Compared to traditional IDs and other machine learning techniques, extensive studies on the general Benchmark dataset suggest that this hybrid approach dramatically increases the accuracy of the detection and reduces false alarm speeds. Serious, architecture is designed for real -time operation, which allows you to identify and address risks as soon as they appear. Thus, a unique R-CNN/LSTM merger architecture is presented to detect real-time infiltration in this work, providing better performance and flexibility on traditional techniques