In the modern period, there is a sharp rise in the number of network-related programs, apps, and services; however, there is a corresponding rise in network security vulnerabilities. Maintaining network security is a challenging but crucial responsibility. Network security requires a process that is capable of recognize to figure out the dangerous activity taking place on it. We refer to this device as an intrusion detection system. To stop breaches, we use conventional network security methods and equipment, including control of entry, both encrypting and firewalls, and anti-virus software, among other things. All of them, however, are unable to successfully defend networks against new attacks. Systems for algorithms to learn (ML) can divide Internet usage into two categories: invasive and regular. This is the first research about the type of deep learning system intended to notice any forms of attack more precisely. The dataset it, or the updated KDD-Cup99, is our subject to machine learning algorithms to evaluate the traits and actions of hostile attackers. In comparison to other methods, random forest performs better in terms of estimated precision and inaccuracy in detection.