Due to the rapid growth of computer system and Internet, network security became crucial issuernfor most organizations. Mostly organizations increase usage of different tools and methods tornsecure their network due to the increase of security threats. Many methods have been developedrnto secure computer networks and communication over the Internet.rnHowever, none of the existing methods developed by different researches have an accuracy ofrndetecting attacks with high detection rate and low false alarm rate. The other thing is most dealrnwith single detection approach with high number of features which is challenging and timernconsuming to implement. Also it will examine only either previously known attacks or unknownrnattacks.rnThis thesis work is devoted to solve those problems using intrusion detection system architecturernthat is based on neural network, signatures and dimension reduction that can promptly detect andrnclassify attacks, whether they are known or never seen before.rnThe proposed hybrid intrusion detection system combines signature based and anomaly basedrntechniques. Signature based open source which uses pattern search for attack detection and thernanomaly based system is developed using machine learning technique. We implementedrndimension reduction using dataset NSL-KDD and train the system using the well knownrnartificial neural network algorithm in the area of intrusion detection.rnThe evaluation of performance and implementation of the proposed hybrid intrusion detectionrnsystem are made with Java programming language using NetBeans. The results obtained by thernimplementation and evaluation are measured in comparison with other works done using singlerndetection approach. The result shows that the output is encouraging and further refinement of thernwork can produce more robust and reliable intrusion detection system.rnKeywords: Hybrid, Intrusion Detection, Anomaly Detection, SNORT, Artificial NeuralrnNetwork, Principal Component Analysis