Date of Publication :31st July 2023

Abstract: The analysis of a company's computer network activity is crucial to early detection and diminution of insider threats, which are of growing concern to many businesses. These extreme harm acts, such intellectual property theft and the publication of sensitive information, are typically carried out by authorised users. An online unsupervised learning in deep learning proceeds towards detect abnormal network activity from logs of system is presented as a prospective filter for human analysts. Restricted Boltzmann Machines (RBMs) have been the subject of some exploratory research as a network intrusion detection strategy. The key difficulty with current technological progress is to recognize the internal danger within the cloud network. Once information is lost, cloud users are harder to compromise. When security and confidentiality aren't guaranteed, cloud computing is unreliable. The likelihood that sensitive applications like banks, hospitals, and companies may be harmed by genuine user threats is higher. An invader is described as a network user and is displayed as a user. After entering into the network as an insider, they will try to attack important information as it is being communicated or exchanged. The network of a cloud has a number of options for external security. The way of identifying an inside assault utilizing AI technology is the main topic of this study. Utilizing nodes of vulnerable user systems makes it feasible for an inside assault. They will connect to the network using a poor user ID, log in, and make the assertion that they are a trusted node. Then, it is quite challenging to recognize them and they may easily hack and assault information as insiders. In our proposed study, we keep an eye on attackers using a deep learning method and a model of user interaction behaviour. User activities from the actual user are recorded in a database. Furthermore, it will notify the system of an insider threat. Analyse the behaviour of individuals within the organization that is malicious. Our main intention from this project is to provide user-friendly software applications. The system detects insider threats, alerts the system, and identifies malicious behaviour within the organization.

Reference :