This study leverages electroencephalogram (EEG) data for the classification of emotions. To enhance the accuracy of the classification process, a multi-objective hyperparameter search is employed, specifically tailored to optimize the performance of the Multi-Layer Perceptron (MLP) classifier. A pivotal aspect of this methodology lies in effectively managing the trade-off between accuracy and computational complexity. This challenge is effectively addressed through the implementation of the Non-dominated Sorting Genetic Algorithm (NSGA-II) as the optimization framework. By adopting this approach, the study offers a robust and efficient solution for emotion classification, showcasing its relevance and effectiveness within the domain of affective computing. Moreover, the findings contribute significantly to the broader understanding of how to navigate the intricate balance between accuracy and computational resource demands in the realm of machine learning applications.