Wang Zhou

Accurate breast cancer (BC) diagnosis is a difficult task that is critical for the proper treatment of BC in IoMT (Internet of Medical Things) healthcare systems. This paper proposes a convolutional neural network (CNN)-based diagnosis method for detecting early-stage breast cancer. In developing the proposed method, we incorporated the CNN model for the invasive ductal carcinoma (IDC) classification using breast histology image data. We have incorporated transfer learning (TL) and data augmentation (DA) mechanisms to improve the CNN model's predictive outcomes. For the fine-tuning process, the CNN model was trained with breast histology image data. Furthermore, the held-out cross-validation method for best model selection and hyper-parameter tuning was incorporated. In addition, various performance evaluation metrics for model performance assessment were computed. The experimental results confirmed that the proposed model outperformed the baseline models across all evaluation metrics, achieving 99.04% accuracy. We recommend the proposed method for early recognition of BC in IoMT healthcare systems due to its high performance.

In this article we have illustrated an efficient and effective neural framework referred to as NFCS for cold start recommendation. In this neural network, the average ratings will be as the input, and the missing ratings will be regarded as the dropout rate for the neural network. Therefore, in the output layer ratings will be reconstructed, which could provide an high performance solution for cold start problem. Experimental results also demonstrate that NFCS could outperform state-of-the-art PMF and IRCD-ICS over real life datasets.