Logo image
Beyond Balance: Addressing class imbalance in fine-tuning deep learners
Journal article   Open access   Peer reviewed

Beyond Balance: Addressing class imbalance in fine-tuning deep learners

Moritz Mock, Thomas Borsani, Giuseppe Di Fatta and Barbara Russo
Science of Computer Programming, Vol.253, pp.1-10
253
2026
Handle:
https://hdl.handle.net/10863/51446

Abstract

Code Comment Classification Imbalanced data Loss-Weighting Deep Learner
Datasets often contain heavily underrepresented classes. Class imbalance biases models toward frequent classes, reducing performance on rare but important categories; in-process strategies such as loss-weighting remain under-explored for software engineering artefacts. We investigate loss-weighting functions for code comment classification and package our methods into Beyond Balance, a reusable implementation offering multiple weighting strategies for Transformer- and Sentence-Transformer–based models. Loss weighting consistently improves F1 performance across datasets, demonstrating an effective and easily adoptable imbalance-handling technique through Beyond Balance.
pdf
1-s2.0-S0167642326000407-main4.02 MBDownloadView
Open Access
url
https://www.sciencedirect.com/science/article/pii/S0167642326000407?via%3DihubView

Details

Metrics

1 Record Views
Logo image