Study for Performance of Un-Pretrained and Pre-trained Models based on CNN

Abstract

In recent years, as the accuracy of deep learning algorithms in image classification tasks exceeds that of the human brain, Artificial Intelligence (AI) auxiliary diagnosis systems have attracted more and more attention. In this paper, some commonly used Convolutional Neural Network (CNN) models e.g. MobileNet, VGG and ResNet are trained and compared on the cancer detection dataset, and it is found that the pre-trained models based on the idea of the transfer learning perform better than the newly trained models in terms of training speed and model performance. Thus, it can be seen that the transfer learning method has great potential in the field of cancer diagnosis. This study provides some experimental support and suggestions on how to further improve the property of the transfer learning method in the field of cancer diagnosis. Meantime, the performance of VGG19 can be proved to be better compared to other models (i.e., MobileNet and ResNet).

Related Papers

• → Implementing convolutional neural network model for prediction in medical imaging(2022)6 cited
• → Leaf Features Extraction for Plant Classification using CNN(2021)7 cited
• → Deep Convolution Neural Network for RBC Images(2022)2 cited
• → CNN-based Transfer Learning for Covid-19 Diagnosis(2021)3 cited
• → Deep Convolutional Neural Networks(2021)8 cited