Abstract
This paper describes an optimization technique for data augmentation (DA) in an acoustic scene classification (ASC) task. DA techniques are necessary to train machine learning models adequately using a small amount of training data. Although various DA techniques have been proposed, the appropriate DA process differs depending on the task. In this paper, we show that the appropriate DA process for an ASC task can be determined using Differentiable Automatic Data Augmentation (DADA). DADA can search for combinations of predefined DA methods and probabilistically select only those methods that effectively improve the accuracy of the ASC model. In our experiments using the ESC-50 dataset, the ASC’s accuracy was improved from the baseline value of 75.8% to 79.3% by applying the appropriate DA process determined using DADA.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.