In the present era, society is highly dependent on technology and digitization. Due to digitization, amount of different kinds of data is increasing exponentially. Organizations and individuals are becoming completely dependent on data which need security. The importance of maintaining confidentiality, integrity, authenticity is becoming an utmost important issue. Intrusion Detection System (IDS) provides a wide range of security for any host, network devices and network. Lot of computer researchers are involved in research activities related to IDS since last few decades. With the technological growth of artificial intelligence, researchers are focusing on the design of IDSs based on machine learning techniques for achieving higher level of efficiency. Data Preprocessing is one of the major contributing techniques to achieve higher outcomes from a machine learning technique. This research paper focuses on achieving the higher level of efficiency of IDS with optimized resources. Dimension reduction is applied for data preprocessing before using machine learning technique for classification. Based on the considered objects after dimension reduction, Neural Network with multilayer perceptron, classification is applied on network traffic to determine if there is an anomaly traffic, and the required action is notified to the system or to the system administrator. The satisfactory level of classification accuracy is achieved for designing IDS with minimum level of resources in this research.