Applying Transfer Learning to Accelerate Cancer Classification and Prediction
TatiReddy Ravi, Shashidhar Gurav, Nandhini Nandhini, Vijayaraj, S. K. Muthusundar · 2025
This chapter describes the use of transfer learning for the classification and prediction of cancer, leveraging the knowledge of the pre-trained InceptionV3 model. After using the Cancer Genome Atlas (TCGA) dataset to extensively preprocess the data and select features to prepare the data for analysis, the model was adapted to provide a new classification layer so that it could be applied to a variety of cancer types. The training process involved data augmentation and the Adam optimizer. Having achieved a training accuracy of 95% and validation accuracy of 92%, the model functioned at a test set accuracy of 91%, precision of 0.88, recall of 0.89, and AUC of 0.94. Having provided evidence of the potential benefits of transfer learning for the accelerated development of solutions for cancer diagnosis, it provides an avenue for more effective clinical use.