Scholarlink

<p>Big data classification involves analyzing data, capturing the data, storage, visualization, and data retrieval. Recent methods suffer from huge time consumption, complexity, imbalance, and overfitting problems. To overcome these limitations, this research proposed an opportunistic scavenging algorithm optimized with a light gradient boosting machine and convolutional neural network (OSA-G2CNN) for data classification, where data balancing is performed using the optimized generative adversarial network (GAN). The proposed method aimed at higher accuracy and faster training efficiency with automated feature extraction that affords major benefits, like risk management, regulatory compliance, and legal discovery. The optimized GAN renders data balancing with enhanced performance and provides rapid convergence with lower computational complexity, which enhances the overall model performance efficacy. Additionally, the model utilizes the concept drift adaptation method, which improves classification accuracy and provides better reliability results in data distribution under a dynamic environment. The achievements of the proposed model in terms of accuracy, specificity, and sensitivity are 96.55%, 96.49%, and 96.60%, respectively, for the lung dataset, and 94.31%, 93.76%, and 94.86%, respectively, for the cervical cancer 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

OSA-G2CNN: Deep Learning-Based Big Data Classification of Medical Images

  • 期刊论文

  • 发表日期：

    2025年10月19日

查看全文
  • Pradnya Bhangale,
  • Pradheep Manisekaran,
  • R. P. Sharma

摘要

Big data classification involves analyzing data, capturing the data, storage, visualization, and data retrieval. Recent methods suffer from huge time consumption, complexity, imbalance, and overfitting problems. To overcome these limitations, this research proposed an opportunistic scavenging algorithm optimized with a light gradient boosting machine and convolutional neural network (OSA-G2CNN) for data classification, where data balancing is performed using the optimized generative adversarial network (GAN). The proposed method aimed at higher accuracy and faster training efficiency with automated feature extraction that affords major benefits, like risk management, regulatory compliance, and legal discovery. The optimized GAN renders data balancing with enhanced performance and provides rapid convergence with lower computational complexity, which enhances the overall model performance efficacy. Additionally, the model utilizes the concept drift adaptation method, which improves classification accuracy and provides better reliability results in data distribution under a dynamic environment. The achievements of the proposed model in terms of accuracy, specificity, and sensitivity are 96.55%, 96.49%, and 96.60%, respectively, for the lung dataset, and 94.31%, 93.76%, and 94.86%, respectively, for the cervical cancer dataset.