arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2106.11629cs.LGcs.SE
收藏

合成代码生成的对抗鲁棒性研究

On Adversarial Robustness of Synthetic Code Generation

Indian Institute of Technology Gandhinagar(甘地纳加尔印度理工学院)

Mrinal Anand, Pratik Kayal, Mayank Singh

摘要:

从自然语言描述自动合成代码是一项具有挑战性的任务。我们见证了近年来在采用序列到序列深度学习技术开发领域特定语言(DSL)的代码生成系统方面取得的巨大进展。在本文中,我们专门对基于\textsc{AlgoLisp} DSL的生成模型进行了实验,并通过不同类别的对抗性示例展示了显著的数据集偏差的存在。我们还试验了基于Transformer的模型的两种变体,它们的性能优于所有现有的基于\textsc{AlgoLisp} DSL的代码生成基线。与当前最先进的系统一致,我们提出的模型在对抗性环境下也表现不佳。因此,我们提出了几种数据集增强技术,以减少偏差,并使用强大的实验来展示其功效。

英文摘要:

Automatic code synthesis from natural language descriptions is a challenging task. We witness massive progress in developing code generation systems for domain-specific languages (DSLs) employing sequence-to-sequence deep learning techniques in the recent past. In this paper, we specifically experiment with \textsc{AlgoLisp} DSL-based generative models and showcase the existence of significant dataset bias through different classes of adversarial examples. We also experiment with two variants of Transformer-based models that outperform all existing \textsc{AlgoLisp} DSL-based code generation baselines. Consistent with the current state-of-the-art systems, our proposed models, too, achieve poor performance under adversarial settings. Therefore, we propose several dataset augmentation techniques to reduce bias and showcase their efficacy using robust experimentation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑