arXiv Analytics

Sign in

arXiv:1712.09662 [cs.CL]AbstractReferencesReviewsResources

CNN Is All You Need

Qiming Chen, Ren Wu

Published 2017-12-27Version 1

The Convolution Neural Network (CNN) has demonstrated the unique advantage in audio, image and text learning; recently it has also challenged Recurrent Neural Networks (RNNs) with long short-term memory cells (LSTM) in sequence-to-sequence learning, since the computations involved in CNN are easily parallelizable whereas those involved in RNN are mostly sequential, leading to a performance bottleneck. However, unlike RNN, the native CNN lacks the history sensitivity required for sequence transformation; therefore enhancing the sequential order awareness, or position-sensitivity, becomes the key to make CNN the general deep learning model. In this work we introduce an extended CNN model with strengthen position-sensitivity, called PoseNet. A notable feature of PoseNet is the asymmetric treatment of position information in the encoder and the decoder. Experiments shows that PoseNet allows us to improve the accuracy of CNN based sequence-to-sequence learning significantly, achieving around 33-36 BLEU scores on the WMT 2014 English-to-German translation task, and around 44-46 BLEU scores on the English-to-French translation task.

Related articles: Most relevant | Search more
arXiv:1606.09370 [cs.CL] (Published 2016-06-30)
Relation extraction from clinical texts using domain invariant convolutional neural network
arXiv:1603.04513 [cs.CL] (Published 2016-03-15)
Multichannel Variable-Size Convolution for Sentence Classification
arXiv:1708.03421 [cs.CL] (Published 2017-08-11)
N-gram and Neural Language Models for Discriminating Similar Languages