Neural Code Comprehension: A Learnable Representation of Code Semantics

TL;DR

This paper introduces inst2vec, a language-independent code embedding learned from an intermediate representation, which effectively captures program semantics and improves performance prediction and algorithm classification tasks.

Contribution

It proposes a novel code embedding space based on IR and contextual flow, enabling robust semantic understanding across programming languages.

Findings

Embeddings outperform specialized approaches in performance prediction.

State-of-the-art results in algorithm classification from raw code.

Effective qualitative analysis of embeddings through analogies and clustering.

Abstract

With the recent success of embeddings in natural language processing, research has been conducted into applying similar methods to code analysis. Most works attempt to process the code directly or use a syntactic tree representation, treating it like sentences written in a natural language. However, none of the existing methods are sufficient to comprehend program semantics robustly, due to structural features such as function calls, branching, and interchangeable order of statements. In this paper, we propose a novel processing technique to learn code semantics, and apply it to a variety of program analysis tasks. In particular, we stipulate that a robust distributional hypothesis of code applies to both human- and machine-generated programs. Following this hypothesis, we define an embedding space, inst2vec, based on an Intermediate Representation (IR) of the code that is independent of the source programming language. We provide a novel definition of contextual flow for this IR, leveraging both the underlying data- and control-flow of the program. We then analyze the embeddings qualitatively using analogies and clustering, and evaluate the learned representation on three different high-level tasks. We show that even without fine-tuning, a single RNN architecture and fixed inst2vec embeddings outperform specialized approaches for performance prediction (compute device mapping, optimal thread coarsening); and algorithm classification from raw code (104 classes), where we set a new state-of-the-art.