Deep Unsupervised Representation Learning for Abnormal Heart Sound Classification


A comparison of computer algorithm paradigms for the audio classification task of normal, mild abnormalities, and moderate/severe abnormalities in phonocardiogram recordings.
Deep Unsupervised Representation Learning for Abnormal Heart Sound Classification

Authors: Shahin Amiriparian, Universitat Augsburg, Augsburg, Bayern, DE, Maximilian Schmitt, Universitat Augsburg, Augsburg, Bayern, DE, Nicholas Cummins, Universitat Augsburg, Augsburg, Bayern, DE, Kun Qian, Universitat Augsburg, Augsburg, Bayern, DE, Fengquan Dong, Shenzhen University General Hospital, Shenzhen, P. R. China, Björn Schuller, Universitat Augsburg, Augsburg, Bayern, DE

Summary: This study presents a comparison of conventional and state-of-the-art deep learning based computer algorithm paradigms for the audio classification task of normal, mild abnormalities, and moderate/severe abnormalities as present in phonocardiogram recordings. In particular, they explore the suitability of deep feature representations as learnt by sequence to sequence autoencoders based on the auDeep toolkit.

View Research -->