Reinforcement Learning and Data-Generation for Syntax-Guided Synthesis

被引:0
|
作者
Parsert, Julian [1 ,2 ]
Polgreen, Elizabeth [2 ]
机构
[1] Univ Oxford, Oxford, England
[2] Univ Edinburgh, Edinburgh, Midlothian, Scotland
基金
奥地利科学基金会; 英国工程与自然科学研究理事会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Program synthesis is the task of automatically generating code based on a specification. In Syntax-Guided Synthesis (SyGuS) this specification is a combination of a syntactic template and a logical formula, and the result is guaranteed to satisfy both. We present a reinforcement-learning guided algorithm for SyGuS which uses Monte-Carlo Tree Search (MCTS) to search the space of candidate solutions. Our algorithm learns policy and value functions which, combined with the upper confidence bound for trees, allow it to balance exploration and exploitation. A common challenge in applying machine learning approaches to syntax-guided synthesis is the scarcity of training data. To address this, we present a method for automatically generating training data for SyGuS based on anti-unification of existing first-order satisfiability problems, which we use to train our MCTS policy. We implement and evaluate this setup and demonstrate that learned policy and value improve the synthesis performance over a baseline by over 26 percentage points in the training and testing sets. Our tool outperforms state-of-the-art tool cvc5 on the training set and performs comparably in terms of the total number of problems solved on the testing set (solving 23% of the benchmarks on which cvc5 fails). We make our data set publicly available, to enable further application of machine learning methods to the SyGuS problem.
引用
收藏
页码:10670 / 10678
页数:9
相关论文
共 50 条
  • [21] Syntax-guided text generation via graph neural network
    Qipeng GUO
    Xipeng QIU
    Xiangyang XUE
    Zheng ZHANG
    ScienceChina(InformationSciences), 2021, 64 (05) : 67 - 76
  • [22] Syntax-guided text generation via graph neural network
    Qipeng Guo
    Xipeng Qiu
    Xiangyang Xue
    Zheng Zhang
    Science China Information Sciences, 2021, 64
  • [23] Syntax-guided text generation via graph neural network
    Guo, Qipeng
    Qiu, Xipeng
    Xue, Xiangyang
    Zhang, Zheng
    SCIENCE CHINA-INFORMATION SCIENCES, 2021, 64 (05)
  • [24] A syntax-guided multi-task learning approach for Turducken-style code generation
    Guang Yang
    Yu Zhou
    Xiang Chen
    Xiangyu Zhang
    Yiran Xu
    Tingting Han
    Taolue Chen
    Empirical Software Engineering, 2023, 28
  • [25] Scalable Algorithms for Abduction via Enumerative Syntax-Guided Synthesis
    Reynolds, Andrew
    Barbosa, Haniel
    Larraz, Daniel
    Tinelli, Cesare
    AUTOMATED REASONING, PT I, 2020, 12166 : 141 - 160
  • [26] A syntax-guided multi-task learning approach for Turducken-style code generation
    Yang, Guang
    Zhou, Yu
    Chen, Xiang
    Zhang, Xiangyu
    Xu, Yiran
    Han, Tingting
    Chen, Taolue
    EMPIRICAL SOFTWARE ENGINEERING, 2023, 28 (06)
  • [27] Syntax-Guided Enumeration of Temporal Properties
    Martino, Gianluca
    Fey, Goerschwin
    PROCEEDINGS OF THE 2019 FORUM ON SPECIFICATION AND DESIGN LANGUAGES (FDL), 2019,
  • [28] Perses: Syntax-Guided Program Reduction
    Sun, Chengnian
    Li, Yuanbo
    Zhang, Qirun
    Gu, Tianxiao
    Su, Zhendong
    PROCEEDINGS 2018 IEEE/ACM 40TH INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING (ICSE), 2018, : 361 - 371
  • [29] Syntax-guided controllable sentence simplification
    Wang, Lulu
    Wumaier, Aishan
    Yibulayin, Tuergen
    Maimaiti, Maihemuti
    NEUROCOMPUTING, 2024, 587
  • [30] Exact and Approximate Methods for Proving Unrealizability of Syntax-Guided Synthesis Problems
    Hu, Qinheping
    Cyphert, John
    D'Antoni, Loris
    Reps, Thomas
    PROCEEDINGS OF THE 41ST ACM SIGPLAN CONFERENCE ON PROGRAMMING LANGUAGE DESIGN AND IMPLEMENTATION (PLDI '20), 2020, : 1128 - 1142