IDS (Intrusion Detection System) is a security component that protects computer and network systems. Variety of methods have been developed to improve the IDS accuracy. One of the most recent approaches is the use of real-time monitoring and irregular activity detection. When an intrusion is detected, a message will be sent to the network administrator. One of the disadvantages of IDS is the possibility of a bad packet by-passing through network traffic. As a result, an improvement on the Artificial Neural Network (ANN) is explored in this study to enhance attack detection in IDS. Standard and attack events are described using the NSL-KDD dataset. In this study, the Magnetic Optimization Algorithm (MOA) is combined with Particle Swarm Optimization (PSO) named PSOMOA, thus to increase the classification rate and achieve high detection ac-curacy in IDS. MOA is a heuristic optimization algorithm which deals with at-traction between particles scattered in the search space and inspired by magnetic field theory in physics. NSL-KDD dataset represented as attacks and normal activities used in this study. Smurf and Neptune attacks are selected for testifying detection and classification ability of attack category of proposed PSOMOA. During the experimentation process, four of the most important features of the dataset were selected. The PSOMOA findings are compared to those of the other form, which employs PSO and MOA. According to the obtained results, the proposed PSOMOA could increase IDS accuracy by up to 99.5 percent.