Impact of noisy labels in learning techniques: A survey

Abstract

Noisy data is the main issue in classification. The possible sources of noise label can be insufficient availability of information or encoding/communication problems, or data entry error by experts/nonexperts, etc., which can deteriorate the model’s performance and accuracy. However, in a real-world dataset, like Flickr, the likelihood of containing the noisy label is high. Initially, few methods such as identification, correcting, and elimination of noisy data was used to enhance the performance. Various machine learning algorithms are used to diminish the noisy environment, but in the recent studies, deep learning models are resolving this issue. In this survey, a brief introduction about the solution for the noisy label is provided. © Springer Nature Singapore Pte Ltd. 2020.