Scholarlink

<p>Presently, the Internet of Things (IoT) is designed with mobile wireless networks and distributed systems, which generate huge amounts of data in the network edge. In various applications, learning-based communication securely manages confidential datasets while sharing insights to train sophisticated machine learning models. This approach analyzes customer satisfaction and uses collective privacy learning to ensure secure training. However, implementing this in a decentralized environment introduces challenges due to significant data variations across different devices and locations. Therefore, a new customer analysis model is developed using federated learning in the IoT environment. Initially, the customer analysis data are collected from a standard RFM analysis dataset to facilitate precise behavioural profiling. Further, the Federated Learning-based Multi-scale Dilated LSTM and Weighted DNN (FL-MDLSTM-WDNN) approach is implemented for analyzing the customer data, where the parameters are optimized using Adaptively Enhanced Wild Geese Migration Optimization (AEWGMO) to improve the performance of the customer analysis model. Finally, the performance of the implemented framework is contrasted with the existing techniques to validate the effectiveness of the proposed customer analysis framework. The RMSE, MAE and MASE values of the developed framework are 15.09, 2.65, and 15.54 respectively. Hence, the solutions confirmed that the recommended customer analysis framework achieves highly satisfying outcomes.</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

Advanced Federated Learning-Based Customer Analysis Model with Improved Wild Geese Migration Optimization for Providing New Products and Services

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月8日

查看全文
  • R. Siva Subramanian,
  • Angeline Lydia,
  • J. Aswini,
  • Sudha Kothandapani

摘要

Presently, the Internet of Things (IoT) is designed with mobile wireless networks and distributed systems, which generate huge amounts of data in the network edge. In various applications, learning-based communication securely manages confidential datasets while sharing insights to train sophisticated machine learning models. This approach analyzes customer satisfaction and uses collective privacy learning to ensure secure training. However, implementing this in a decentralized environment introduces challenges due to significant data variations across different devices and locations. Therefore, a new customer analysis model is developed using federated learning in the IoT environment. Initially, the customer analysis data are collected from a standard RFM analysis dataset to facilitate precise behavioural profiling. Further, the Federated Learning-based Multi-scale Dilated LSTM and Weighted DNN (FL-MDLSTM-WDNN) approach is implemented for analyzing the customer data, where the parameters are optimized using Adaptively Enhanced Wild Geese Migration Optimization (AEWGMO) to improve the performance of the customer analysis model. Finally, the performance of the implemented framework is contrasted with the existing techniques to validate the effectiveness of the proposed customer analysis framework. The RMSE, MAE and MASE values of the developed framework are 15.09, 2.65, and 15.54 respectively. Hence, the solutions confirmed that the recommended customer analysis framework achieves highly satisfying outcomes.