A Comparative Analysis Using Transfer Learning

Abstract

Through little adjustments, Transfer Learning enables us to leverage pre-trained models from other individuals. We will explain in this article how we may leverage pre-trained models to speed up our solutions. It is a well-liked deep learning strategy that has been effectively used in numerous areas and has recently drawn more attention from researchers. The growth in certain sectors is restricts when it becomes challenging to obtain a large-scale, well-annotated dataset because of the expense collecting the data.  Transfer learning is a machine learning technique that involves applying a model created for one job to another task. In deep learning, transfer learning is gaining popularity because it allows deep neural networks to be trained with less data than if a model had to be created from the beginning. The assumption that the training and test sets should have the same distribution and be independent is lessened by transfer learning. The problem regarding insufficient training data is resolved through transfer learning. The examination of modern deep neural network-based transfer learning research and its applications is the main goal of this survey. We defined deep transfer learning, classified recent research papers using deep transfer learning approaches, and reviewed them.