Repulsive Attention: Rethinking Multi-head Attention as Bayesian Inference

by   Bang An, et al.

The neural attention mechanism plays an important role in many natural language processing applications. In particular, the use of multi-head attention extends single-head attention by allowing a model to jointly attend information from different perspectives. Without explicit constraining, however, multi-head attention may suffer from attention collapse, an issue that makes different heads extract similar attentive features, thus limiting the model's representation power. In this paper, for the first time, we provide a novel understanding of multi-head attention from a Bayesian perspective. Based on the recently developed particle-optimization sampling techniques, we propose a non-parametric approach that explicitly improves the repulsiveness in multi-head attention and consequently strengthens model's expressiveness. Remarkably, our Bayesian interpretation provides theoretical inspirations on the not-well-understood questions: why and how one uses multi-head attention. Extensive experiments on various attention models and applications demonstrate that the proposed repulsive attention can improve the learned feature diversity, leading to more informative representations with consistent performance improvement on various tasks.


page 5

page 18

page 19

page 20

page 21

page 22


Orthogonality Constrained Multi-Head Attention For Keyword Spotting

Multi-head attention mechanism is capable of learning various representa...

Multi-head or Single-head? An Empirical Comparison for Transformer Training

Multi-head attention plays a crucial role in the recent success of Trans...

Cascaded Head-colliding Attention

Transformers have advanced the field of natural language processing (NLP...

What Do Recurrent Neural Network Grammars Learn About Syntax?

Recurrent neural network grammars (RNNG) are a recently proposed probabi...

Are Sixteen Heads Really Better than One?

Attention is a powerful and ubiquitous mechanism for allowing neural mod...

Multi-Head Attention with Disagreement Regularization

Multi-head attention is appealing for the ability to jointly attend to i...

Low-Rank Bottleneck in Multi-head Attention Models

Attention based Transformer architecture has enabled significant advance...