Enhancing Sentiment Analysis Accuracy through Data driven Hyperparameter Optimization Using Convolutional Neural Network

Every dataset has its unique kind of data and individual pieces of information. Certain configurations exhibit promising results across diverse datasets. This paper discusses the identification of optimal hyperparameters for sentiment analysis through systematic experimentation. In this implementation, Convolutional Neural Network (CNN) is used for sentiment analysis by interpreting facial expressions to identify seven emotions: happy, sad, neutral, angry, fearful, disgusted, and surprised. However, these methods can still get better because sometimes don’t work perfectly due to various reasons like how quickly the algorithm learns and how much data it processes at once. Our objective is to enhance the accuracy of sentiment analysis models by identifying hyperparameter configurations that serve as effective starting points, acknowledging that these initial settings may not be perfect but can provide a solid foundation for subsequent refinement.