@inproceedings{2ab0bb7dd61f48abaedf66ef64d0a000,
title = "Sinif dengesizliǧi probleminde {\c c}eki{\c s}meli {\"u}retici aǧlarin kullanimi",
abstract = "Having more samples belonging to one class than the samples of the other class in data used in a classification task is known as class imbalance problem. Handling class imbalance is crucial since the classifier's performance is highly affected. One of the solution approaches of this problem is to make the data balanced by generating synthetic data. Employing resampling methods is a common way of generating synthetic data. Although adversarial generative networks (GANs) are mainly designed to generate image data, they can also be an alternative to solve the class imbalance problem by generating tabular data. This work presents a comparative study of resampling methods with GANs based methods. The performance of machine learning methods improved by 27% if the data is balanced with resampling methods. However, similar performance results were observed with working on imbalanced data if the GANs based methods are employed for synthetic data generation.",
keywords = "Class imbalance problem, Generative adversarial networks, Resampling",
author = "Aydin, {M. Asli}",
note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 29th IEEE Conference on Signal Processing and Communications Applications, SIU 2021 ; Conference date: 09-06-2021 Through 11-06-2021",
year = "2021",
month = jun,
day = "9",
doi = "10.1109/SIU53274.2021.9477939",
language = "T{\"u}rk{\c c}e",
series = "SIU 2021 - 29th IEEE Conference on Signal Processing and Communications Applications, Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "SIU 2021 - 29th IEEE Conference on Signal Processing and Communications Applications, Proceedings",
}