MétaCan
Menu
Back to cohort
Record W4390679819 · doi:10.1109/dsc59305.2023.00078

FixGPT: A Novel Three-Tier Deep Learning Model for Automated Program Repair

2023· article· en· W4390679819 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

Venuenot available
Typearticle
Languageen
FieldComputer Science
TopicSoftware Testing and Debugging Techniques
Canadian institutionsBrock University
Fundersnot available
KeywordsComputer scienceIdentifierArtificial intelligenceMachine learningDeep learningBenchmarkingTransformerData miningProgramming language

Abstract

fetched live from OpenAlex

Automatic program repair plays a crucial role in the software development and implementation. While deep learning-based approaches have made significant progress, one inherent challenge is the inefficiency in code representation, which hampers accurate patch generation. Furthermore, the training data used by these data-driven approaches may be limited, and they may not be able to capture the subtle differences between vulnerabilities and patches. To address these issues, FixGPT, we propose a three-tier deep learning model in the study. Specifically, a generative pre-trained transformer model is designed in the first tier to capture code characteristics and programming patterns. The second tier integrates a generation model based on the structure of neural machine translation, for the purpose of generating potential patches. Finally, a contrast model is introduced in the last tier to differentiate between the vulnerability and the patch. We also incorporate the Byte Pair Encoding approach to reduce the search space by converting identifiers into subwords. Detailed experimental studies have been carried out to evaluate the performance of FixGPT on two well-known benchmarking datasets: QuixBugs and Defects4J. The results demonstrated significant improvements in the effectiveness and accuracy in comparison with existing solutions. We complement these findings through the analysis of two case studies.

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.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.001
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Methods · Consensus signal: Methods
Teacher disagreement score0.900
Threshold uncertainty score0.515

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.001
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.001
Science and technology studies0.0000.000
Scholarly communication0.0000.000
Open science0.0010.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.054
GPT teacher head0.324
Teacher spread0.271 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it