eduzhai > Applied Sciences > Engineering >

Deep Dense and Convolutional Autoencoders for Unsupervised Anomaly Detection in Machine Condition Sounds

  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: This technical report describes two methods that were developed for Task 2 ofthe DCASE 2020 challenge. The challenge involves an unsupervised learning todetect anomalous sounds, thus only normal machine working condition samples areavailable during the training process. The two methods involve deepautoencoders, based on dense and convolutional architectures that usemelspectogram processed sound features. Experiments were held, using the sixmachine type datasets of the challenge. Overall, competitive results wereachieved by the proposed dense and convolutional AE, outperforming the baselinechallenge method.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...