Szczegóły publikacji
Opis bibliograficzny
AttentionMix: a guided text data augmentation method relying on attention / Dominik Lewy, Jacek MAŃDZIUK // W: Neural Information Processing : 31st International Conference, ICONIP 2024 : Auckland, New Zealand, December 2–6, 2024 : proceedings , Pt. 9 / eds. Mufti Mahmud, [et al.]. — Singapore: Springer Nature, cop. 2025. — ( Lecture Notes in Computer Science ; ISSN 0302-9743 ; vol. 15294 ). — ISBN: 978-981-96-6601-0; e-ISBN: 978-981-96-6599-0. — S. 409–423. — Bibliogr., Abstr. — Publikacja dostępna online od: 2025-07-02. — J. Mańdziuk - dod. afiliacja: Warsaw University of Technology, Warsaw
Autorzy (2)
- Lewy Dominik
- AGHMańdziuk Jacek
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 163080 |
|---|---|
| Data dodania do BaDAP | 2025-09-30 |
| DOI | 10.1007/978-981-96-6599-0_28 |
| Rok publikacji | 2025 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Springer |
| Konferencja | International Conference on Neural Information Processing 2024 |
| Czasopismo/seria | Lecture Notes in Computer Science |
Abstract
The Mixup method has proven to be a powerful data augmentation technique in Computer Vision, with many successors that perform image mixing in a guided manner. One of the interesting research directions is transferring the underlying Mixup idea to other domains, e.g. Natural Language Processing (NLP). Even though there already exist several methods that apply Mixup to textual data, there is still room for new, improved approaches. In this work, we introduce AttentionMix, a novel mixing method that relies on attention-based information. While the paper focuses on the BERT attention mechanism, the proposed approach can technically be applied with any attention-based model. AttentionMix is evaluated on 3 standard sentiment classification datasets and in most of the tested setups outperforms two benchmark approaches that utilize Mixup mechanism, as well as the vanilla BERT method. These findings demonstrate that attention-based information can be effectively used for data augmentation in the NLP domain, indicating the relevance of further research along this line.
