Cross-Language Learning for Program Classification using BTBCNNs

Towards the vision of automatically translating code that implements an algorithm from one programming language into another, this paper proposes an approach for automated program classifications using bilateral tree-based convolutional neural networks (BiTBCNNs). It is layered on top of two tree-based convolutional neural networks (TBCNNs), each of which recognizes the algorithm of code written in an individual programming language. The combination layer of the networks recognizes the similarities and differences among code in different programming languages. The BiTBCNNs are trained using the source code in different languages but known to implement the same algorithms and/or functionalities. For a preliminary evaluation, we use 3591 Java and 3534 C++ code snippets from 6 algorithms we crawled systematically from GitHub. We obtained 90+% accuracy in the cross-language binary classification task to tell whether any given two code snippets implement a same algorithm. Also, for the actual algorithm classification task, i.e., to predict the algorithm label of an arbitrary C++ code snippet, we achieve 80.5% for the precision. Therefore, the capability of BiTBCNNs for classifying programs into algorithms across programming languages may beome a useful building block towards automated program translation.