Macformer: Transformer with Random Maclaurin Feature Attention

TL;DR

Macformer introduces a novel Transformer architecture using Random Maclaurin features to efficiently approximate dot-product kernels, significantly accelerating attention computations for long sequences while maintaining accuracy.

Contribution

It proposes Macformer, a new Transformer model employing RMF for kernel approximation, with a novel regularization mechanism, improving efficiency and accuracy for long sequence processing.

Findings

Macformer accelerates attention computation on long sequences.

RMFA provides an unbiased approximation of kernelized attention.

Experimental results align with theoretical analysis, confirming effectiveness.

Abstract

Random feature attention (RFA) adopts random fourier feature (RFF) methods to approximate the softmax function, resulting in a linear time and space attention mechanism that enables the construction of an efficient Transformer. Inspired by RFA, we propose Macformer, a Transformer architecture that employs random Maclaurin features (RMF) to approximate various dot-product kernels, thereby accelerating attention computations for long sequence. Macformer consists of Random Maclaurin Feature Attention (RMFA) and pre-post Scaling Batch Normalization (ppSBN), the former is an unbiased approximation for dot-product kernelized attention and the later is a two-stage regularization mechanism guaranteeing the error of RMFA. We conducted toy experiments to demonstrate the efficiency of RMFA and ppSBN, and experiments on long range arena (LRA) benchmark to validate the acceleration and accuracy of Macformer with different dot-product kernels. Experiment results of Macformer are consistent with our theoretical analysis.