Machine Learning2024
Fuzzy-Hybrid Emotion Classification
MSc Dissertation · University of the West of England, Bristol
Developed a novel fuzzy-hybrid approach combining BERT transformers, LSTM networks, and Support Vector Machines with lexicon-based methods for emotion classification. Implemented ANFIS to handle uncertainty in emotional states using Plutchik’s 8-emotion model. Applied RFE for optimal feature selection and compared performance against Bi-LSTM with attention mechanisms to classify complex emotional expressions beyond traditional sentiment analysis.
BERTLSTMSVMANFISFuzzy logicRFEPlutchik