Programming by Examples: PL Meets ML

摘要

Programming by Examples (PBE) involves synthesizing intended programs in an underlying domain-specific language from example-based specifications. PBE systems are already revolutionizing the application domain of data wrangling and are set to significantly impact several other domains including code refactoring. There are three key components in a PBE system, (i) A search algorithm that can efficiently search for programs that are consistent with the examples provided by the user. We leverage a divide-and-conquer-based deductive search paradigm that inductively reduces the problem of synthesizing a program expression of a certain kind that satisfies a given specification into sub-problems that refer to sub-expressions or sub-specifications, (ii) Program ranking techniques to pick an intended program from among the many that satisfy the examples provided by the user. We leverage features of the program structure as well of the outputs generated by the program on test inputs, (iii) User interaction models to facilitate usability and debuggability. We leverage active-learning techniques based on clustering inputs and synthesizing multiple programs. Each of these PBE components leverage both symbolic reasoning and heuristics. We make the case for synthesizing these heuristics from training data using appropriate machine learning methods. This can not only lead to better heuristics, but can also enable easier development, maintenance, and even personalization of a PBE system.