This paper extends our previous work by presenting the Enhanced Multi Layer Perceptron(MLP) model as a method for intrusion detection and classification, as we seek to combat a growing challenge of securing networks against emerging cyber threats. Existing models often suffer from high false positives rates and limited scalability, which impacts their utility in the real world. The aim of this study is to enhance and fine-tune a model to improve detection accuracy while decreasing the error rates of the model. The Enhanced MLP was evaluated through extensive experimentation using a real-world dataset in order to allow for an evaluation of many different categories of attacks and network behavior anomalies. Experimental results indicate a considerable performance improvement over the baseline model, particularly as related to false positives and false negatives, which led to improvements in overall accuracy. Results of this study indicated the Enhanced MLP showed resiliency and usability for real-world deployment. In summary, we conclude the Enhanced MLP improves automated threat detection and can provide insight into the future development of next generation intrusion detection systems (IDS).