Scholarlink

<p>This study presents a novel approach to text-based emotion recognition using dual-branch Convolutional Neural Network (CNN) architecture. Our objectives were to address class imbalance in emotion datasets and improve classification accuracy across multiple emotion categories. We innovated by implementing a balanced sampling technique and a unique dual-branch CNN design for feature extraction. The models achieved 99.59% accuracy on the test set, with precision and recall both exceeding 99.5%. While demonstrating high performance, limitations include potential over fitting and the need for larger, more diverse datasets. This research contributes to advancing emotion recognition in text, with potential applications in sentiment analysis, customer service, and human–computer interaction.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A deep learning approach for text-based emotion recognition: improving accuracy through dual-branch CNN architecture and balanced sampling

  • 期刊论文

  • 发表日期：

    2025年10月4日

查看全文
  • M. Asha Priyadarshini,
  • A. Krishna Mohan

摘要

This study presents a novel approach to text-based emotion recognition using dual-branch Convolutional Neural Network (CNN) architecture. Our objectives were to address class imbalance in emotion datasets and improve classification accuracy across multiple emotion categories. We innovated by implementing a balanced sampling technique and a unique dual-branch CNN design for feature extraction. The models achieved 99.59% accuracy on the test set, with precision and recall both exceeding 99.5%. While demonstrating high performance, limitations include potential over fitting and the need for larger, more diverse datasets. This research contributes to advancing emotion recognition in text, with potential applications in sentiment analysis, customer service, and human–computer interaction.