Scholarlink

<p>With the rapid growth of the Internet of Things (IoT), a massive volume of sensory data is continuously generated, creating challenges in storage, processing and privacy, particularly in medical applications. This research hypothesizes that integrating optimization techniques with deep learning (DL) in a federated learning (FL) framework can enhance the accuracy and efficiency of big data classification while preserving privacy. Therefore, a Fossa Skill Optimization-enabled Deep Q Net in Federated Learning (FL-FSO_DQN) is introduced in this paper, which is the integration of FL, Fossa Skill Optimization (FSO) and Deep Q Net (DQN). FL is utilized for training the model with better privacy by maintaining local models on devices and aggregating updates on a global server. During local training, the features from input data are selected by Bray Curtis distance and Random Variable (RV) coefficient. The DQN classifies the big data, and the FSO trains its parameters for improved performance. The implementation of FL-FSO_DQN is performed with Python tool employing COVID-19 Machine Learning Dataset, Covid19PatientsAnalysis and Heart disease dataset. Experimental evaluation demonstrates that the proposed approach attains the accuracy, True Positive Rate (TPR), True Negative Rate (TNR), Mean Squared Error (MSE), Root MSE (RMSE), and loss of 93.46%, 93.95%, 92.57%, 0.133, 0.182, and 0.110.</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

Fossa Skill Optimization Enabled Deep Learning for Medical Big Data Classification in IoT Using Federated Learning

  • 期刊论文

  • 发表日期：

    2026年1月17日

查看全文
  • M. Nandini,
  • Deepak N. Biradar

摘要

With the rapid growth of the Internet of Things (IoT), a massive volume of sensory data is continuously generated, creating challenges in storage, processing and privacy, particularly in medical applications. This research hypothesizes that integrating optimization techniques with deep learning (DL) in a federated learning (FL) framework can enhance the accuracy and efficiency of big data classification while preserving privacy. Therefore, a Fossa Skill Optimization-enabled Deep Q Net in Federated Learning (FL-FSO_DQN) is introduced in this paper, which is the integration of FL, Fossa Skill Optimization (FSO) and Deep Q Net (DQN). FL is utilized for training the model with better privacy by maintaining local models on devices and aggregating updates on a global server. During local training, the features from input data are selected by Bray Curtis distance and Random Variable (RV) coefficient. The DQN classifies the big data, and the FSO trains its parameters for improved performance. The implementation of FL-FSO_DQN is performed with Python tool employing COVID-19 Machine Learning Dataset, Covid19PatientsAnalysis and Heart disease dataset. Experimental evaluation demonstrates that the proposed approach attains the accuracy, True Positive Rate (TPR), True Negative Rate (TNR), Mean Squared Error (MSE), Root MSE (RMSE), and loss of 93.46%, 93.95%, 92.57%, 0.133, 0.182, and 0.110.