Different Transfer Learning Approaches for Recognition of Lung Sounds: Review

Lung disorders are currently the leading cause of mortality in the world. The majority of instances of lung illness are discovered after the condition worsens to an advanced level; however, early diagnosis of lung disease may assist in the treatment of the condition. In today’s world, technology innovations play an essential part in the delivery of healthcare services. The aim of this study in the healthcare arena is focused on the necessity for diagnosing lung disease via lung sounds, which is a current research subject. In the medical system, transfer learning is essential for success. This review provides a variety of Transfer learning techniques for the categorization of lung sounds, including ALEXNET, MOBILENET, VGGNET, and RESNET, amongst others, in this paper. The classification will also be performed using the Modified RESNET with Mel spectrogram of lung sound signals in the Transfer learning model, in addition to the aforementioned methods. These transfer learning models exhibit great accuracy in lung sound categorization, and they have the potential to be utilized in the future to identify lung disorders. This review will address several Transfer Learning Approaches, as well as their advantages and disadvantages. There are other disadvantages. In order to recognize four different classes of lung sounds. Additionally, provide future study directions on the recognition of lung sounds.