Beyond Self-Attention: External Attention Using Two Linear Layers for Visual Tasks

🤖 Plain-English Summary

Attention mechanisms, especially self-attention, have played an increasingly important role in deep feature representation for visual tasks. We further incorporate the multi-head mechanism into external attention to provide an all-MLP architecture, external attention MLP (EAMLP), for image classification.

🔑 Key Findings

• Self-attention updates the feature at each position by computing a weighted sum of features using pair-wise affinities across all positions to capture the long-range dependency within a single sample.
• However, self-attention has quadratic complexity and ignores potential correlation between different samples.
• This article proposes a novel attention mechanism which we call external attention, based on two external, small, learnable, shared memories, which can be implemented easily by simply using two cascaded linear layers and two normalization layers; it conveniently replaces self-attention in existing popular architectures.

🔗 Related Resources