Details
Poster
Presenter(s)
Display Name
Qinxin Zhou
- Affiliation
-
AffiliationUniversities of Electronic Science and Technology of China
- Country
Abstract
This paper proposed a training sample balancing method for SVEB accuracy enhancement on the MIT-BIH arrythmias database. Wavelet features extracted from the class balanced ECG data are applied to a simple one-layer CNN-based model. Despite a low computation complexity for the used algorithm, the proposed method gives a 10% improvement for SVEB detection rate comparing with other AI-assisted algorithms that also utilize MOE training scheme.