阶段监督的潜在推理用于单次JavaScript反混淆
Stage-Supervised Latent Reasoning for Single-Shot JavaScript Deobfuscation
- The Pennsylvania State University（宾夕法尼亚州立大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对现有LLM反混淆方法忽略阶段结构的问题，提出阶段感知潜在推理框架，利用Coconut训练从多阶段重写学习，推理时单次生成，在JsDeObsBench上语法有效性达50%，语义正确性达80%。
AI中文摘要：
JavaScript混淆被广泛用于保护代码，但它也使程序分析和安全审查变得相当困难。现有的基于LLM的反混淆方法通常将任务视为一步翻译，忽略了实际反混淆流水线的阶段结构。这篇WIP论文提出了一个阶段感知的潜在推理框架，将确定性反混淆工具的中间输出转换为基于Coconut的训练监督。模型在训练期间从多阶段重写中学习，但在推理时以单次生成最终清理后的程序。在JsDeObsBench上的初步结果显示，基于Coconut的模型将语法有效性提高到50%，而直接微调为15%，零样本基线为25%，并且在有效输出中达到80%的语义正确性，表明在反混淆中比两个比较模型具有更好的语义保真度。
英文摘要：
JavaScript obfuscation is widely used to protect code, but it also makes program analysis and security review substantially harder. Existing LLM-based deobfuscation methods usually treat the task as one-step translation, ignoring the staged structure of practical deobfuscation pipelines. This WIP paper proposes a stage-aware latent reasoning framework that converts intermediate outputs from a deterministic deobfuscation tool into supervision for Coconut-based training. The model learns from multi-stage rewrites during training but generates the final cleaned program in a single shot at inference time. Preliminary results on JsDeObsBench show that the Coconut-based model improves syntactic validity to 50%, compared with 15% for direct fine-tuning and 25% for a zero-shot baseline, and reaches 80% semantic correctness among valid outputs, indicating better semantic faithfulness than either comparison model in deobfuscation.
