Intrusion Detection Systems (IDS) in Software-Defined Networking (SDN) play an important role in improving network security by monitoring and analyzing traffic. The centralized control of SDN allows IDS to detect and respond to threats more effectively and adaptively in real time. Traditional networks often face challenges in handling complex security issues due to limited monitoring capabilities. This study applies Ant Colony Optimization (ACO) for feature selection, which uses pheromone-based exploration to identify the best features and improve model accuracy. Metaheuristic algorithms like ACO solve complex optimization problems using iterative, nature-inspired techniques. Several machine learning models, including Logistic Regression, Naive Bayes, Linear Discriminant Analysis, Adaline, and Random Forest, were tested on a standard dataset for intrusion detection. Among these, Random Forest showed the highest accuracy of 99.41%, making it the most reliable for detecting network threats. This research highlights the benefits of combining ACO-based feature selection with machine learning classifiers, providing insights into developing stronger IDS to address evolving cybersecurity risks.