Scholarlink

<p>The human-computer interaction analyzes how people utilize technology and how it can learn from people. Brain computer interfaces are utilized by experts to understand how technology may be used to assist people’s behavior and intelligence. Electroencephalography (EEG) is a harmless, portable, affordable, well-known, and user-friendly system for measuring brain waves. Humans are affected by several mental and physical health issues due to the stress incurred by a busy lifestyle. It impacts the quality of life since it poses several mental disorders. People have practiced yoga and meditation to enjoy a stress-free life. EEG signal analysis can be used for analyzing the effects of yoga and meditation, especially in human brain. But, it takes more time and effort to gather more detailed information from unprocessed EEG data for investigating the negative impacts on the human being. Although research on the beneficial effects of meditation on controlling emotions is increasing,&#xa0;the effects of meditation on regulating emotions have not been explored in previous research. Thus, an intellectual EEG-based human brain activity interpretation mechanism during meditation and yoga is implemented in this research work. At first, from the available data sources, the EEG signals are garnered, and then the artifacts presented in the original EEG signals are removed. After that, the artifact-removed signals are effectively given into the feature extraction stage, whereas the wave features, Linear Discriminant Analysis (LDA) features, and statistical features are extracted. Moreover, these extracted features are concatenated and then, the features are optimally selected by the Mutated Random Parameter-based Mother Optimization Algorithm (MRP-MOA). Henceforth, the optimal selected features are given to the Adaptive Hybrid Network (AHN), where the Capsule Network (CapsNet) is integrated with the Extreme Learning Machine (ELM) for analyzing human brain activity. Here, the same MRP-MOA is supported to optimally tune the AHN network parameters. At last, the experiments are performed for the designed approach to examine the model’s effectiveness by comparing them with traditional techniques. On the analysis, the proposed model achieves a high accuracy value, rather than 6.68% of CNN, 6.56% of LSTM, 5.34% of ELM and 6.25% of CapsNet, respectively.</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

An Optimal Feature Integration with Adaptive Hybrid Deep Learning Network for Human Brain Activity Interpretation During Yoga and Meditation Using EEG Signals

  • 期刊论文

  • 发表日期：

    2026年3月26日

查看全文
  • Bandari Ramaraju,
  • Ravichander Janapati,
  • Sreedhar Kollem

摘要

The human-computer interaction analyzes how people utilize technology and how it can learn from people. Brain computer interfaces are utilized by experts to understand how technology may be used to assist people’s behavior and intelligence. Electroencephalography (EEG) is a harmless, portable, affordable, well-known, and user-friendly system for measuring brain waves. Humans are affected by several mental and physical health issues due to the stress incurred by a busy lifestyle. It impacts the quality of life since it poses several mental disorders. People have practiced yoga and meditation to enjoy a stress-free life. EEG signal analysis can be used for analyzing the effects of yoga and meditation, especially in human brain. But, it takes more time and effort to gather more detailed information from unprocessed EEG data for investigating the negative impacts on the human being. Although research on the beneficial effects of meditation on controlling emotions is increasing, the effects of meditation on regulating emotions have not been explored in previous research. Thus, an intellectual EEG-based human brain activity interpretation mechanism during meditation and yoga is implemented in this research work. At first, from the available data sources, the EEG signals are garnered, and then the artifacts presented in the original EEG signals are removed. After that, the artifact-removed signals are effectively given into the feature extraction stage, whereas the wave features, Linear Discriminant Analysis (LDA) features, and statistical features are extracted. Moreover, these extracted features are concatenated and then, the features are optimally selected by the Mutated Random Parameter-based Mother Optimization Algorithm (MRP-MOA). Henceforth, the optimal selected features are given to the Adaptive Hybrid Network (AHN), where the Capsule Network (CapsNet) is integrated with the Extreme Learning Machine (ELM) for analyzing human brain activity. Here, the same MRP-MOA is supported to optimally tune the AHN network parameters. At last, the experiments are performed for the designed approach to examine the model’s effectiveness by comparing them with traditional techniques. On the analysis, the proposed model achieves a high accuracy value, rather than 6.68% of CNN, 6.56% of LSTM, 5.34% of ELM and 6.25% of CapsNet, respectively.