Scholarlink

COVID-19 transmission is still a huge hidden danger for elderly people with underlying diseases and immunocompromised infants and children, COVID-19 is a kind of viral pneumonia, and it is important to accurately and quickly distinguish COVID-19 from common viral pneumonia. Recently, the Red-Billed Blue Magpie Algorithm (RBMO), a high-performance optimiser, has achieved very significant performance in model parameter optimisation. In order to take advantage of the great potential of RBMO, we propose a new classification method for the task of classifying CT images of pneumonia, called RBMO-Att-Bi-LSTM. This method is the first time that RBMO has been applied to bi-directional long and short-term memory networks (Bi-LSTM), where RBMO can optimise the parameters such as the number of nodes in the hidden layer of the model, the regularisation coefficients, and the learning rate, and at the same time, it uses the self-attentive mechanism to capture the dependencies at different locations in the input sequence. Compared with other state-of-the-art methods, the experimental results show that RBMO-Att-Bi-LSTM has the best performance in classifying CT images of pneumonia and proves the effectiveness of RBMO-Att-Bi-LSTM.

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

RBMO-Att-Bi-LSTM: A Red-Billed Blue Magpie Optimiser-Self-attention Mechanism Based Optimisation of Bi-Directional Long- and Short-Term Memory Networks for Classification of COVID-19 CT Images

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yihao Sun,
  • Tianming Du,
  • Hongzan Sun,
  • Jian Xu,
  • Md Mamunur Rahaman,
  • Xinghao Wang,
  • Xinyu Huang,
  • Tao Jiang,
  • Marcin Grzegorzek,
  • Ning Xu,
  • Chen Li

摘要

COVID-19 transmission is still a huge hidden danger for elderly people with underlying diseases and immunocompromised infants and children, COVID-19 is a kind of viral pneumonia, and it is important to accurately and quickly distinguish COVID-19 from common viral pneumonia. Recently, the Red-Billed Blue Magpie Algorithm (RBMO), a high-performance optimiser, has achieved very significant performance in model parameter optimisation. In order to take advantage of the great potential of RBMO, we propose a new classification method for the task of classifying CT images of pneumonia, called RBMO-Att-Bi-LSTM. This method is the first time that RBMO has been applied to bi-directional long and short-term memory networks (Bi-LSTM), where RBMO can optimise the parameters such as the number of nodes in the hidden layer of the model, the regularisation coefficients, and the learning rate, and at the same time, it uses the self-attentive mechanism to capture the dependencies at different locations in the input sequence. Compared with other state-of-the-art methods, the experimental results show that RBMO-Att-Bi-LSTM has the best performance in classifying CT images of pneumonia and proves the effectiveness of RBMO-Att-Bi-LSTM.