Projects
Facial Expression Recognition leveraging MobileNet architecture, k-fold testing, and data augmentation techniques to improve dataset balance, achieving accuracy of 75.71% on Google FER 2013 data. The project aims to improve human-computer interaction by enabling systems to respond to emotional cues in real-time applications.
Conducted a thorough analysis of a comprehensive dataset obtained from Fitbit users to uncover key trends and insights. The objective was to identify significant patterns in user behavior, understand their implications for customer engagement, and leverage these insights to shape effective marketing strategies.
Aspect-Based Sentiment Analysis
Implemented a sophisticated Multi-Layer Dual Attention Deep Learning model, utilizing refined word embeddings to evaluate customer reviews on various products and services. This system accurately discerns and categorizes sentiments, providing valuable insights into specific aspects of user experiences, thus enabling informed decision-making and enhancing customer satisfaction
