SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair

2026-04-18

AI Summary

This work addresses the challenge that large language models often struggle to simultaneously preserve semantic correctness and syntactic compliance when automatically repairing code vulnerabilities. The authors propose a novel neuro-symbolic framework that mimics developers’ alternating use of pattern matching and tool feedback by employing a neural routing model to dynamically select between supervised fine-tuning (SFT) and compiler-driven symbolic reward-based fine-tuning (RFT). This adaptive training strategy effectively integrates code synthesis with compiler feedback, achieving substantial improvements over existing methods on the FixJS and CodeFlaws benchmarks. Specifically, the approach yields up to an 18% increase in CodeBLEU and CrystalBLEU scores and a 32% improvement in Exact Match accuracy.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Neuro-Symbolic Learning
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Large Language Models (LLMs) show promise for automated code repair but often struggle with the complex semantic and structural correctness required. We present SynthFix, a hybrid neural-symbolic framework that improves LLM-based vulnerability repair by unifying code synthesis with compiler-informed symbolic feedback. The core of our approach is an adaptive training strategy where a neural Router Model directs code samples to either Supervised Fine-Tuning (SFT) to learn common patterns or Reward Fine-Tuning (RFT) with symbolic rewards for complex, iterative refinement. On the FixJS (JavaScript) and CodeFlaws (C) benchmarks, SynthFix achieves up to 18% relative improvement in CodeBLEU/CrystalBLEU and 32% in Exact Match over strong SFT and RFT baselines. Our results show that this adaptive combination of training strategies, which mirrors how developers alternate between pattern application and tool feedback, significantly improves the accuracy and efficiency of LLM-based vulnerability repair. Our code and data are available at https://github.com/CoderDoge1108/SynthFix.

Problem

• code vulnerability repair
• semantic correctness
• structural correctness
• large language models

Innovation

• neuro-symbolic
• adaptive training
• code vulnerability repair
• symbolic feedback
• large language models