Review on Deep Learning based Network Security Tools in Detecting Real-Time Vulnerabilities

Abstract

Network connected hardware and software systems are always open to vulnerabilities when they are connected with an outdated firewall or an unknown Wi-Fi access. Therefore network based anti-virus software and intrusion detection systems are widely installed in every network connected hardwares. However, the pre-installed security softwares are not quite capable in identifying the attacks when evolved. Similarly, the traditional network security tools that are available in the current market are not efficient in handling the attacks when the system is connected with a cloud environment or IoT network. Hence, recent algorithms of security tools are incorporated with the deep learning network for improving its intrusion detection rate. The adaptability of deep learning network is comparatively high over the traditional software tools when it is employed with a feedback network. The feedback connections included in the deep learning networks produce a response signal to their own network connections as a training signal for improving their work performances. This improves the performances of deep learning-based security tools while it is in real-time operation. The motive of the work is to review and present the attainments of the deep learning-based vulnerability detection models along with their limitations.