Timezone: »

 
Poster
Neural Program Generation Modulo Static Analysis
Rohan Mukherjee · Yeming Wen · Dipak Chaudhari · Thomas Reps · Swarat Chaudhuri · Christopher Jermaine

Tue Dec 07 08:30 AM -- 10:00 AM (PST) @
in Poster Session 1 »

State-of-the-art neural models of source code tend to be evaluated on the generation of individual expressions and lines of code, and commonly fail on long-horizon tasks such as the generation of entire method bodies. We propose to address this deficiency using weak supervision from a static program analyzer. Our neurosymbolic method allows a deep generative model to symbolically compute, using calls to a static analysis tool, long-distance semantic relationships in the code that it has already generated. During training, the model observes these relationships and learns to generate programs conditioned on them. We apply our approach to the problem of generating entire Java methods given the remainder of the class that contains the method. Our experiments show that the approach substantially outperforms a state-of-the-art transformer and a model that explicitly tries to learn program semantics on this task, both in terms of producing programs free of basic semantic errors and in terms of syntactically matching the ground truth.

Keywords: Deep Learning · Transformers · Generative Model

Author Information

Rohan Mukherjee (Rice University)
Yeming Wen (University of Texas, Austin)
Dipak Chaudhari (The University of Texas at Austin)
Thomas Reps (University of Wisconsin-Madison)
Swarat Chaudhuri (The University of Texas at Austin)
Christopher Jermaine (Rice University)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors