Scholarlink

<p>Emotion recognition holds significant importance in mental health assessment and human–computer interaction, with electroencephalogram (EEG) signals emerging as a key research focus due to their ability to directly reflect brain activity characteristics. This study targets university students as the research subjects, addressing challenges in EEG-based emotion recognition such as high-dimensional feature data, noticeable noise, significant temporal shifts, and incomplete feature extraction dimensions. We propose an intelligent emotion recognition system integrating machine learning and clustering algorithms. To improve the performance of emotion recognition, a research proposes an intelligent emotion recognition system that integrates machine learning and clustering algorithms, which is designed from the aspects of feature selection, transfer learning, and data processing. First, the minimum redundancy maximum correlation (mRMR) algorithm and multi-source domain adaptive selection algorithm are used for high discriminative feature screening. Next, an improved K-means clustering algorithm was applied to dynamically allocate weights and select the optimal migration group. Finally, the optimal transfer features are input into a hybrid neural network, and emotion classification and prediction are achieved by convolving generalized features through a hybrid graph. The results demonstrate that the designed method achieved an accuracy of over 95% in convergence validation, and the recognition accuracy of different EEG signal features exceeded 80%. The average processing time for emotional features was less than 1.5&#xa0;s, and emotion recognition accuracy exceeded 85%. The system performs exceptionally well in processing EEG signals, maintaining an emotion recognition accuracy rate above 85% with stable feature dimensions. Through algorithm optimization and system implementation, this study not only effectively mitigates the limitations of EEG-based emotion recognition but also contributes to enhancing the research potential and innovation in the field of emotion applications.</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

Emotion recognition intelligent system based on machine learning and clustering algorithm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月18日

查看全文
  • Xingxing Ge

摘要

Emotion recognition holds significant importance in mental health assessment and human–computer interaction, with electroencephalogram (EEG) signals emerging as a key research focus due to their ability to directly reflect brain activity characteristics. This study targets university students as the research subjects, addressing challenges in EEG-based emotion recognition such as high-dimensional feature data, noticeable noise, significant temporal shifts, and incomplete feature extraction dimensions. We propose an intelligent emotion recognition system integrating machine learning and clustering algorithms. To improve the performance of emotion recognition, a research proposes an intelligent emotion recognition system that integrates machine learning and clustering algorithms, which is designed from the aspects of feature selection, transfer learning, and data processing. First, the minimum redundancy maximum correlation (mRMR) algorithm and multi-source domain adaptive selection algorithm are used for high discriminative feature screening. Next, an improved K-means clustering algorithm was applied to dynamically allocate weights and select the optimal migration group. Finally, the optimal transfer features are input into a hybrid neural network, and emotion classification and prediction are achieved by convolving generalized features through a hybrid graph. The results demonstrate that the designed method achieved an accuracy of over 95% in convergence validation, and the recognition accuracy of different EEG signal features exceeded 80%. The average processing time for emotional features was less than 1.5 s, and emotion recognition accuracy exceeded 85%. The system performs exceptionally well in processing EEG signals, maintaining an emotion recognition accuracy rate above 85% with stable feature dimensions. Through algorithm optimization and system implementation, this study not only effectively mitigates the limitations of EEG-based emotion recognition but also contributes to enhancing the research potential and innovation in the field of emotion applications.