Classification And Detection Of Covid-19 Using Deep Transfer Learning Techniques

The goal of this project is to create the first high-transfer environment for usable analytics. This environment uses AI techniques to solve an application problem in a new domain to train an artificial intelligence model that will deliver state-of the-art performance.Transfer learning can be defined as the ability to take knowledge from outside a domain and apply it to the current domain. This is most commonly used in machine learning when we want to transfer knowledge from one problem solving scenario (a source domain) to another (a target domain). Transfer learning is especially useful for applied machine learning because it allows us to focus on one domain (the target) and use past experience to solve new problems. Recent work in the field of transfer learning has primarily focused on taking knowledge learned from an image, image captioning problem, or using a combination of pre-built machine learning libraries to solve a specific problem. This work has looked at classifying the existence of COVID19 in a person using CT-scans and chest X-rays. In the past, transfer learning often ignored the challenge of improving a certain model as it takes longer to fine tune the features in the dataset. We are taking a different approach where we are using transfer learning to improve on certain features. While the image classification work has spurred an explosion of new techniques, there are important issues that have been left unresolved. Chief among them is the fact that existing techniques require large amounts of data to train on. This is not acceptable for many applications who may only have small amounts of data available to use for training. When training models on small datasets, it can be difficult to ensure that your model will scale to larger datasets.