In real-life scenarios, there is always a possibility of certain classes being underrepresented. The second reason is oversampling of the minority classes. It is highly possible that working directly on the database containing the original samples can be misused or breached. The first reason for creating synthetic samples is to avoid the usage of original samples for privacy reasons. The creation of synthetic data for minority classes can be useful for many reasons. Variational Autoencoders have been used to generate synthetic samples and improve the performance of the classifiers. Standard data augmentation techniques like random translations, flips, and rotations, and the addition of Gaussian noise generates authentic but very little data. Data Augmentation techniques enhance the generalizability of the deep learning model by using the original data efficiently. In the case of a low data scheme, the parameters are underdetermined, and the networks generalize poorly. Deep Learning can be used to unravel the clinically pertinent information from the dataset provided by hospitals, which can be then used for decision making, treatment, and prevention of health conditions.įor efficacious training of neural networks, sufficient data is required. The increase in access to data generated by health care and the development of deep learning techniques has proved to be prosperous. Deep learning is rapidly developing in the field of acoustics, providing many compelling solutions to problems like Automatic Speech Recognition, sound synthesis, acoustic scene classification, generative music, acoustic event detection, and many more. The inception of deep learning has paved the way for many breakthroughs in science, medicine, and engineering. The funder had played no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.Ĭompeting interests: The authors have declared that no competing interests exist. įunding: Symbiosis International (Deemed University) has provided the financial support for the manuscript APC and infrastructure support for implementing the proposed work. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.ĭata Availability: Data can be accessed through the following link. Received: AugAccepted: MaPublished: August 12, 2022Ĭopyright: © 2022 Saldanha et al. PLoS ONE 17(8):Įditor: Murugappan M., Kuwait College of Science and Technology, KUWAIT Hence, our work shows that deep learning-based lung sound classification models are not only a promising solution over traditional methods but can also achieve a significant performance boost upon augmenting an imbalanced training set.Ĭitation: Saldanha J, Chakraborty S, Patil S, Kotecha K, Kumar S, Nayyar A (2022) Data augmentation using Variational Autoencoders for improvement of respiratory disease classification. A significant improvement in the classification performance metrics was observed upon augmenting the imbalanced dataset for certain minority classes and marginal improvement for the other classes. Our results showed that MLP-VAE achieved an average FAD of 12.42 over all classes, whereas Convolutional VAE and Conditional CVAE achieved an average FAD of 11.58 and 11.64 for all classes, respectively. We evaluated the quality of the synthetic respiratory sounds’ quality using metrics such as Fréchet Audio Distance (FAD), Cross-Correlation and Mel Cepstral Distortion. This work aims to synthesize respiratory sounds of various categories using variants of Variational Autoencoders like Multilayer Perceptron VAE (MLP-VAE), Convolutional VAE (CVAE) Conditional VAE and compare the influence of augmenting the imbalanced dataset on the performance of various lung sound classification models. The publicly available ICBHI respiratory sounds database is severely imbalanced, making it difficult for a deep learning model to generalize and provide reliable results. Computerized auscultation of lung sounds is gaining importance today with the availability of lung sounds and its potential in overcoming the limitations of traditional diagnosis methods for respiratory diseases.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |