Program synthesis promises to dramatically revolutionize and democratize the act of programming by lowering the barriers of expressing desired program specifications. While we have made tremendous progress in improving the scalability of search-based synthesis approaches, the current techniques still can not generate programs of relatively modest complexity. On the other hand, human programmers routinely write programs of varying complexity without the need of performing expensive search. In this talk, I will present some of our recent works that take inspiration from this remarkable reasoning ability of humans to develop new program synthesis techniques. In particular, we develop neuro-symbolic techniques that explore the ideas of learning high-level abstractions of primitives using a curriculum of synthesis tasks of increasing complexity, understanding multi-modal specifications, and using program execution based feedback to iteratively improve the learning process.

Rishabh Singh is a research scientist at Google Brain working on neural program synthesis. His research interests span the areas of programming languages and deep learning. Previously, he was a researcher at Microsoft Research. He obtained his PhD in Computer Science from MIT in 2014, where he was a Microsoft Research PhD fellow and was awarded the MIT’s George M. Sprowls Award for Best PhD Dissertation in Computer Science. He obtained his BTech in Computer Science from IIT Kharagpur in 2008, where he was awarded the Institute Silver Medal and Bigyan Sinha Memorial Award.