首页|Tianjin University Reports Findings in Machine Learning (Learning With Incomplet e Labels of Multisource Datasets for Ecg Classification)
Tianjin University Reports Findings in Machine Learning (Learning With Incomplet e Labels of Multisource Datasets for Ecg Classification)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – Research findings on Machine Learning are discussed in a new report. According to news reporting originating in Tianji n, People’s Republic of China, by NewsRx journalists, research stated, “The shor tage of annotated ECG data presents a significant impediment, hampering the over all generalization capabilities of machine learning models tailored for automate d ECG classification. The collective integration of multisource datasets present s a potential remedy for this challenge.” Financial support for this research came from National Natural Science Foundatio n of China (NSFC). The news reporters obtained a quote from the research from Tianjin University, “ However, it is crucial to underscore that the mere addition of supplementary dat a does not automatically guarantee performance enhancement, given the unresolved challenges associated with multisource data. In this research, we address one s uch challenge, namely, the issue of incomplete labels arising from the diversity of annotations within multi -source ECG datasets. First, we identified three di stinct types of label missing: dataset-related label missing, supertype missing, and subtype missing. To address the supertype missing effectively, we introduce a novel approach known as offline category mapping which leverages the hierarch ical relationships inherent within the categories to recover the missing superty pe labels. Additionally, two complementary strategies, referred to as prediction masking and online category mapping, are proposed to mitigating the adverse eff ects of subtype and dataset-related label missing on model optimization. These s trategies enhance the model’s ability to identify missing subtypes under conditi ons of weak supervision. These pioneering methodologies are integrated into a de ep learning -based framework designed for multilabel ECG classification. The per formance of our proposed framework is rigorously evaluated using realistic multi -source datasets obtained from the PhysioNet/CinC challenge 2020/2021. The prop osed learning framework exhibits a notable improvement in macro -average precisi on, surpassing the corresponding baseline model by more than 25 % on the test datasets.”
TianjinPeople’s Republic of ChinaAsi aCyborgsEmerging TechnologiesMachine LearningTianjin University
NETL
NSTL
