Pervasive Attention: 2D Convolutional Neural Networks for Sequence-to-Sequence Prediction

TL;DR

This paper introduces a novel 2D convolutional neural network architecture for sequence-to-sequence prediction, replacing traditional encoder-decoder models with attention mechanisms, resulting in improved performance and simplicity.

Contribution

The paper presents a new convolutional approach that integrates attention-like properties throughout the network, outperforming existing encoder-decoder systems in machine translation.

Findings

Outperforms state-of-the-art encoder-decoder models

Simpler architecture with fewer parameters

Demonstrates effective attention-like behavior throughout

Abstract

Current state-of-the-art machine translation systems are based on encoder-decoder architectures, that first encode the input sequence, and then generate an output sequence based on the input encoding. Both are interfaced with an attention mechanism that recombines a fixed encoding of the source tokens based on the decoder state. We propose an alternative approach which instead relies on a single 2D convolutional neural network across both sequences. Each layer of our network re-codes source tokens on the basis of the output sequence produced so far. Attention-like properties are therefore pervasive throughout the network. Our model yields excellent results, outperforming state-of-the-art encoder-decoder systems, while being conceptually simpler and having fewer parameters.