Data mining is now used by many institutions widely and generally. Intrusion detection for network operators & security specialists is one of the top priorities and challenges. Sensitive data, anonymity and device availability from attacks are protected by the Intrusion detection system. In order to describe resources from those in the database through a network, IDS uses data mining techniques. A robust algorithm must also be built to produce successful rules for the detection of attacks. In this paper, optimization algorithms focused on classification were used to detect attacks over the NSL KDD dataset. Depending on this stranglehold, the current method is explained an improved Hoeffiding Induction Tree algorithm to resolve the drawbacks. The results demonstrate that the proposed HITNB algorithm has improved precision, a lower alarm rate and the ability to detect a new type efficiently.