Szczegóły publikacji
Opis bibliograficzny
A comprehensive review of data augmentation methods for support business processes using machine learning and deep learning / Agata Kozina, Michał PIKUS // W: Emerging challenges in intelligent management information systems : proceedings of 28th European Conference on Artificial Intelligence ECAI 2025 - IMIS Workshop : [Bologna, Italy, October 25–30, 2025] , Vol. 2 / eds. Marcin Hernes, Ewa Walaszczyk, Artur Rot. — Cham : Springer Nature Switzerland, cop. 2026. — ( Lecture Notes in Networks and Systems ; ISSN 2367-3370 ; vol. 1643 ). — ISBN: 978-3-032-06610-7; e-ISBN: 978-3-032-06611-4. — S. 66–79. — Bibliogr., Abstr. — Publikacja dostępna online od: 2025-10-12
Autorzy (2)
- Kozina Agata
- AGHPikus Michał
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 163872 |
|---|---|
| Data dodania do BaDAP | 2025-11-21 |
| DOI | 10.1007/978-3-032-06611-4_6 |
| Rok publikacji | 2026 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Springer |
| Konferencja | European Conference on Artificial Intelligence 2025 |
| Czasopismo/seria | Lecture Notes in Networks and Systems |
Abstract
Data augmentation is a critical technique for overcoming data scarcity and imbalances in many machine learning and deep learning applications. This review presents a systematic analysis of data augmentation methods with a focus on their application in decision support systems within management, economics, and finance. Terminological discrepancies between data transformation and augmentation were also taken into account, emphasizing the importance of properly distinguishing between these concepts for the accurate interpretation of the applied methods. Both traditional and advanced augmentation strategies are discussed, including geometric transformations, generative models (GANs, cGANs, VAEs), oversampling techniques such as SMOTE and ADASYN, and the Mixup approach, highlighting their benefits in improving model generalization and accuracy. Challenges related to automation and reproducibility are also addressed, as well as suggestions for future research directions. Additionally, the review explores how classical machine learning models. Especially in structured data contexts -benefit from augmentation techniques such as SMOTE, bootstrapping, and noise injection.
