A Study on Deep Learning Techniques for Emotion Classification

Emotion classification has become a major concern in recent years. Because emotion affects both mental health and physical health. Accurate and early detection of mental stress is very important for timely intervention and mental well-being. Emotion data collected from social media has been widely used for emotion classification. Social media platforms like Reddit and Twitter provide text data for emotion research. In recent years, many studies have been carried out on stress classification using social media data. However, there is still a challenge in capturing semantic and aspect-level correlations, utilizing dependencies, and improving efficiency to develop domain-specific emotion detection systems. Unlike traditional machine learning methods, the advanced deep learning techniques based on transformer utilizes self-attention mechanism to effectively capture complex dependencies within social media data. This study presents related works in deep learning techniques for mental health analysis, stress classification and provides insights into challenges, applications, and emotion classification methods.