Naslov
Synthesizing credit data using autoencoders and generative adversarial networks

Autori
Oreski, Goran

Izvornik
Knowledge-based systems (0950-7051) 274 (2023); 110646, 12

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
Autoencoders ; Generative Adversarial Networks ; Tabular data ; Credit risk data ;

Sažetak
Data quality is an essential element necessary for the development of a successful machine-learning project. One of the biggest challenges in various real-world application domains is class imbalance. This paper proposes a new framework for oversampling credit data by combining two deep learning techniques: autoencoders and generative adversarial networks. A trivial autoencoder (TAE) is used to change data representation, and modied generative adversarial networks (GAN) are used to create new instances from random noise. The experiment on three dierent datasets demonstrates that the same classier achieves a better area under the receiver operating characteristic curve (AUC) on datasets augmented by the proposed framework compared to datasets oversampled by other techniques. Additionally, the results show that datasets balanced by the new framework inuence the classier to change the prediction error types, signicantly reducing false negatives ; more expensive misclassication case in the imbalance learning. The improvements are signicant, and considering the change in error distribution, the proposed technique is an excellent complement to existing oversampling techniques.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo, Informacijske i komunikacijske znanosti

POVEZANOST RADA