ICML-2016 A Convolutional Attention Network for Extreme Summarization of Source Code

· ICML· · convolutional-neural-network, code-summarization

Paper: A Convolutional Attention Network for Extreme Summarization of Source Code

Feature extraction with convolutional attention

The authors argue that translation-invariant structure is common across many domains, and therefore use CNNs to learn representations.

avatar