Protecting industry infrastructure from cyber-attacks is the necessity of time. We heard of zero-day attacks, and now every industry is facing them. The death of the Perimeter term in cybersecurity defines that we can't make our infrastructure 100% secure. We are using Firewalls, Anti-viruses, IPS (Intrusion Prevention Systems) and IDS (Intrusion Detection Systems), etc., to make industry perimeter secure. But we are still facing zero-day attacks, and we can't prevent them. There is a need for smart IDS that can detect intrusion in networks and systems, and prevent them as well. Some previous work is already been done to detect zero-day attacks in networks, and they are using different approaches based on anomaly detection using Machine Learning algorithms. Machine Learning provides many benefits including and not limited to zero-day attacks by analyzing previously learned and trained datasets. But the only problem that remains is the accuracy of ML algorithms. Our approach is using old-fashioned signature-based detection, which guarantees 100% accuracy but is limited to zero-day attacks. To deal with zero-day attacks, we're using our approach with the ML algorithms and a multi-classifier to detect zero-day intrusion in a network and update their signatures automatically. Our approach will guarantee prevention from known threats using signatures and more accuracy in the detection of zero-day attacks by using ML algorithms with a multi-classifier.