R2Vul: Learning to Reason about Software Vulnerabilities with Reinforcement Learning and Structured Reasoning Distillation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Weyssow, Martin, Yang, Chengran, Chen, Junkai, Widyasari, Ratnadira, Zhang, Ting, Huang, Huihui, Nguyen, Huu Hung, Tun, Yan Naing, Bui, Tan, Li, Yikun, Wei, Ang Han, Liauw, Frank, Ouh, Eng Lieh, Shar, Lwin Khin, Lo, David
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918116306452480
author Weyssow, Martin
Yang, Chengran
Chen, Junkai
Widyasari, Ratnadira
Zhang, Ting
Huang, Huihui
Nguyen, Huu Hung
Tun, Yan Naing
Bui, Tan
Li, Yikun
Wei, Ang Han
Liauw, Frank
Ouh, Eng Lieh
Shar, Lwin Khin
Lo, David
author_facet Weyssow, Martin
Yang, Chengran
Chen, Junkai
Widyasari, Ratnadira
Zhang, Ting
Huang, Huihui
Nguyen, Huu Hung
Tun, Yan Naing
Bui, Tan
Li, Yikun
Wei, Ang Han
Liauw, Frank
Ouh, Eng Lieh
Shar, Lwin Khin
Lo, David
contents Large language models (LLMs) have shown promising performance in software vulnerability detection, yet their reasoning capabilities remain unreliable. We propose R2Vul, a method that combines reinforcement learning from AI feedback (RLAIF) and structured reasoning distillation to teach small code LLMs to detect vulnerabilities while generating security-aware explanations. Unlike prior chain-of-thought and instruction tuning approaches, R2Vul rewards well-founded over deceptively plausible vulnerability explanations through RLAIF, which results in more precise detection and high-quality reasoning generation. To support RLAIF, we construct the first multilingual preference dataset for vulnerability detection, comprising 18,000 high-quality samples in C\#, JavaScript, Java, Python, and C. We evaluate R2Vul across five programming languages and against four static analysis tools, eight state-of-the-art LLM-based baselines, and various fine-tuning approaches. Our results demonstrate that a 1.5B R2Vul model exceeds the performance of its 32B teacher model and leading commercial LLMs such as Claude-4-Opus. Furthermore, we introduce a lightweight calibration step that reduces false positive rates under varying imbalanced data distributions. Finally, through qualitative analysis, we show that both LLM and human evaluators consistently rank R2Vul model's reasoning higher than other reasoning-based baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2504_04699
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle R2Vul: Learning to Reason about Software Vulnerabilities with Reinforcement Learning and Structured Reasoning Distillation
Weyssow, Martin
Yang, Chengran
Chen, Junkai
Widyasari, Ratnadira
Zhang, Ting
Huang, Huihui
Nguyen, Huu Hung
Tun, Yan Naing
Bui, Tan
Li, Yikun
Wei, Ang Han
Liauw, Frank
Ouh, Eng Lieh
Shar, Lwin Khin
Lo, David
Software Engineering
Artificial Intelligence
Computation and Language
Large language models (LLMs) have shown promising performance in software vulnerability detection, yet their reasoning capabilities remain unreliable. We propose R2Vul, a method that combines reinforcement learning from AI feedback (RLAIF) and structured reasoning distillation to teach small code LLMs to detect vulnerabilities while generating security-aware explanations. Unlike prior chain-of-thought and instruction tuning approaches, R2Vul rewards well-founded over deceptively plausible vulnerability explanations through RLAIF, which results in more precise detection and high-quality reasoning generation. To support RLAIF, we construct the first multilingual preference dataset for vulnerability detection, comprising 18,000 high-quality samples in C\#, JavaScript, Java, Python, and C. We evaluate R2Vul across five programming languages and against four static analysis tools, eight state-of-the-art LLM-based baselines, and various fine-tuning approaches. Our results demonstrate that a 1.5B R2Vul model exceeds the performance of its 32B teacher model and leading commercial LLMs such as Claude-4-Opus. Furthermore, we introduce a lightweight calibration step that reduces false positive rates under varying imbalanced data distributions. Finally, through qualitative analysis, we show that both LLM and human evaluators consistently rank R2Vul model's reasoning higher than other reasoning-based baselines.
title R2Vul: Learning to Reason about Software Vulnerabilities with Reinforcement Learning and Structured Reasoning Distillation
topic Software Engineering
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2504.04699