Just-in-Time and Real-Time Bug-Inducing Commit Prediction Using a Federated Learning Approach
Bibliographic record
Abstract
Previous studies proposed different methods for predicting Just in Time (JIT) and Real-Time (RT) bug-inducing commits to reduce maintenance costs. While Machine Learning (ML) models have been used to predict bug-inducing commits, a large and robust dataset is required to train the models. Unfortunately, it is difficult for individuals to collect such large datasets due to privacy concerns when sharing data, especially from software projects across multiple organizations. In this regard, during model training, Federated Learning (FL) has been introduced to train collectively to overcome data-sharing limitations due to privacy concerns. In this paper, we apply FL to predict JIT and RT bug-inducing commits and to understand if FL can be a viable solution to overcome the privacy limitations in this domain. In this study, we compare a few standalone ML models, such as Logistic Regression (LR) and Deep Learning (DL), with their respective FL models, namely Federated Logistic Regression (FL-LR) and Federated Deep Learning (FL-DL) for JIT and RT bug-inducing commit prediction. We also compare different aggregation strategies like FedAvg and FedAvgM for FL. The study uses 126,103 commits from 22 projects across three benchmark datasets from various domains. Our study suggests that the federated approach enhances model performance for cross-project commits and reduces training time. The FL models maintain consistency across diverse application domains. Furthermore, aggregation strategies like FedAvgM are beneficial to improve model accuracy for JIT and RT bug-inducing commit prediction.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".