arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2305.18342cs.LGcs.AIcs.CLcs.CYcs.PL
收藏

视觉编程的神经任务合成

Neural Task Synthesis for Visual Programming

Max Planck Institute for Software Systems(马克斯·普朗克软件系统研究所)

Victor-Alexandru Pădurean, Georgios Tzannetos, Adish Singla

AI总结:

本文提出神经符号方法NeurTaskSyn,结合模仿学习与强化学习,自动合成符合指定编程概念和视觉约束的视觉编程任务,并在Code.org和CodeHS的参考任务上验证了有效性。

摘要:

生成式神经模型通过合成新内容,在增强编程教育方面具有巨大潜力。我们致力于设计能够在视觉编程领域背景下,根据给定规范自动生成编程任务的神经模型。尽管GPT-4等大型生成模型近期取得了成功,但我们的初步结果表明,这些模型在合成视觉编程任务方面效果不佳,并且在逻辑推理和空间推理方面存在困难。我们提出了一种新颖的神经符号技术NeurTaskSyn,该技术能够根据以期望的编程概念(由其解代码所运用)和对视觉任务的约束形式给出的规范来合成编程任务。NeurTaskSyn包含两个组件:第一个组件通过模仿学习过程进行训练,以生成可能的解代码;第二个组件通过强化学习过程进行训练,以引导底层符号执行引擎为这些代码生成视觉任务。我们通过广泛的实证评估以及对参考任务的定性研究,证明了NeurTaskSyn的有效性,这些参考任务分别来自Code.org的《编程一小时:经典迷宫》挑战和CodeHS.com的《Karel编程入门》课程。

英文摘要:

Generative neural models hold great promise in enhancing programming education by synthesizing new content. We seek to design neural models that can automatically generate programming tasks for a given specification in the context of visual programming domains. Despite the recent successes of large generative models like GPT-4, our initial results show that these models are ineffective in synthesizing visual programming tasks and struggle with logical and spatial reasoning. We propose a novel neuro-symbolic technique, NeurTaskSyn, that can synthesize programming tasks for a specification given in the form of desired programming concepts exercised by its solution code and constraints on the visual task. NeurTaskSyn has two components: the first component is trained via imitation learning procedure to generate possible solution codes, and the second component is trained via reinforcement learning procedure to guide an underlying symbolic execution engine that generates visual tasks for these codes. We demonstrate the effectiveness of NeurTaskSyn through an extensive empirical evaluation and a qualitative study on reference tasks taken from the Hour of Code: Classic Maze challenge by Code-dot-org and the Intro to Programming with Karel course by CodeHS-dot-com.

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。

↑