REAL

Ensemble noisy label detection on MNIST

Fazekas, István and Barta, Attila and Fórián, László (2021) Ensemble noisy label detection on MNIST. Annales Mathematicae et Informaticae, 53. pp. 125-137. ISSN 1787-6117

[img]
Preview
Text
AMI_53_from125to137.pdf - Published Version

Download (805kB) | Preview

Abstract

In this paper machine learning methods are studied for classification data containing some misleading items. We use ensembles of known noise correction methods for preprocessing the training set. Preprocessing can be either relabeling or deleting items detected to have noisy labels. After preprocessing, usual convolutional networks are applied to the data. With preprocessing, the performance of very accurate convolutional networks can be further improved.

Item Type: Article
Uncontrolled Keywords: Label noise, deep learning, classification
Subjects: Q Science / természettudomány > QA Mathematics / matematika > QA75 Electronic computers. Computer science / számítástechnika, számítógéptudomány
Depositing User: Tibor Gál
Date Deposited: 25 May 2021 09:00
Last Modified: 03 Apr 2023 07:15
URI: http://real.mtak.hu/id/eprint/125739

Actions (login required)

Edit Item Edit Item