Scholarlink

<p>The health care (HC) data encompasses a wide range of information, from patient records to wearable device data, all contributing to better healthcare outcomes and innovations. HC data classification is a crucial process for managing and protecting sensitive information. However, striking a balance between privacy preservation and maintaining useful information is challenging in the previous researches. Here, LeNet Kronecker network (Le-KNet) is designed for privacy protected data classification. Firstly, input HC data is acquired and then, noise fully homomorphic encryption (FHE) is conducted by optimum multi-key generation. The optimal multi-key is generated using Chronological Drawer Puzzle Algorithm (CDPA), which is introduced by combining chronological drawer algorithm (CDA) with Puzzle Optimization algorithm (POA). Furthermore, CDA is integration of chronological concept with Drawer Algorithm (DA). After that, data is decrypted by multi-key to conduct classification of privacy protected data. The classification procedure is executed by Le-KNet that is designed by integrating LeNet with Deep Kronecker network (DKN). The proposed model is compared with the conventional techniques using heart disease datasets including Hungary, Cleveland, VA Long Beach, and Switzerland. With 90% of training data, the proposed Le-KNet model produced the maximum accuracy, true negative rate (TNR), and true positive rate (TPR) of 92.8, 93.1, and 91.5%, respectively, using the VA Long Beach dataset.</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

Chronological Drawer puzzle algorithm based multi-key generation and Lenet Kronecker network enabled privacy protected data classification

  • 期刊论文

  • 发表日期：

    2025年2月25日

查看全文
  • Hemant Ramdas Kumbhar,
  • S. Srinivasa Rao

摘要

The health care (HC) data encompasses a wide range of information, from patient records to wearable device data, all contributing to better healthcare outcomes and innovations. HC data classification is a crucial process for managing and protecting sensitive information. However, striking a balance between privacy preservation and maintaining useful information is challenging in the previous researches. Here, LeNet Kronecker network (Le-KNet) is designed for privacy protected data classification. Firstly, input HC data is acquired and then, noise fully homomorphic encryption (FHE) is conducted by optimum multi-key generation. The optimal multi-key is generated using Chronological Drawer Puzzle Algorithm (CDPA), which is introduced by combining chronological drawer algorithm (CDA) with Puzzle Optimization algorithm (POA). Furthermore, CDA is integration of chronological concept with Drawer Algorithm (DA). After that, data is decrypted by multi-key to conduct classification of privacy protected data. The classification procedure is executed by Le-KNet that is designed by integrating LeNet with Deep Kronecker network (DKN). The proposed model is compared with the conventional techniques using heart disease datasets including Hungary, Cleveland, VA Long Beach, and Switzerland. With 90% of training data, the proposed Le-KNet model produced the maximum accuracy, true negative rate (TNR), and true positive rate (TPR) of 92.8, 93.1, and 91.5%, respectively, using the VA Long Beach dataset.