Review of unknown attack detection with deep learning techniques

Zero-day attacks, which are also known as unknown attacks, are a major threat to computer networks because they take advantage of weaknesses that no one knew about before. Researchers have looked into using convolutional neural networks (CNN) to find zero-day threats and stop them. The use of deep learning methods to the detection of emerging threats in the realm of network security is rapidly growing in importance. The goal of this study is to come up with a general way for creating and training a convolutional neural network (CNN) model for identifying unexpected threats using the KDD Cup 1999 and BoT IoT datasets. To use the suggested method, first prepare the data, then extract features, make a CNN design, train and test models, and then release them. The method could make breach detection systems more effective and efficient and help protect computer networks from security risks. This would be a very good thing. In addition, This paper gives an overview of current study on detecting zero-day attacks with CNN, including methods for collecting and preparing data, CNN structures, training and testing strategies, and evaluation measures. The poll shows the pros and cons of using CNN to find zero-day attacks and points out key research holes and goals for the future. But this method needs more research to figure out how to deal with its limitations and problems and to see if it works in real-world network settings.