Comparison of Deep Learning Evaluation Metrics

Evaluating deep learning or machine learning algorithms is a crucial part of the research work. We may get satisfying results using, say Accuracy score(probabilistic domain) but may perform poorly in Root Mean Square Error(RMSE) evaluation metric. Here, we are gonna use a tumor detection deep learning model as a reference to judge our evaluation metrics. There are multiple deep network segmentation models namely URsD, UIncp, UVgg, and URsEn, used after pre-processing of biomedical scans or image datasets. The segmentation results are finally evaluated to account for the similarity between the actual output and the predicted value with the help of the coefficient of performance indices.