Targeted minority-class augmentation with BERT-based contextual word insertion improves fine-grained food hazard classification for BERT by about 6%, but gains are inconsistent across techniques and categories.
A.4 Transformer Models Details In this section, we explain the encoder-only trans- former models’ details and architectures we used in the experiments
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
BrightCookies at SemEval-2025 Task 9: Exploring Data Augmentation for Food Hazard Classification
Targeted minority-class augmentation with BERT-based contextual word insertion improves fine-grained food hazard classification for BERT by about 6%, but gains are inconsistent across techniques and categories.