Showing 1–1 of 1 results for author: Colt, R

Automatic Feature Extraction for Heartbeat Anomaly Detection

Authors: Robert-George Colt, Csongor-Huba Várady, Riccardo Volpi, Luigi Malagò

Abstract: We focus on automatic feature extraction for raw audio heartbeat sounds, aimed at anomaly detection applications in healthcare. We learn features with the help of an autoencoder composed by a 1D non-causal convolutional encoder and a WaveNet decoder trained with a modified objective based on variational inference, employing the Maximum Mean Discrepancy (MMD). Moreover we model the latent distribut… ▽ More We focus on automatic feature extraction for raw audio heartbeat sounds, aimed at anomaly detection applications in healthcare. We learn features with the help of an autoencoder composed by a 1D non-causal convolutional encoder and a WaveNet decoder trained with a modified objective based on variational inference, employing the Maximum Mean Discrepancy (MMD). Moreover we model the latent distribution using a Gaussian chain graphical model to capture temporal correlations which characterize the encoded signals. After training the autoencoder on the reconstruction task in a unsupervised manner, we test the significance of the learned latent representations by training an SVM to predict anomalies. We evaluate the methods on a problem proposed by the PASCAL Classifying Heart Sounds Challenge and we compare with results in the literature. △ Less

Submitted 24 February, 2021; originally announced February 2021.

Comments: 7 pages, 2 figures, Presented at PharML 2020 Workshop - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD), see https://sites.google.com/view/pharml2020/accepted-submissions, source-code: https://github.com/rist-ro/argo

MSC Class: 68T07