DeepAI AI Chat
Log In Sign Up

Linear Self-Attention Approximation via Trainable Feedforward Kernel

by   Uladzislau Yorsh, et al.
Czech Technical University in Prague

In pursuit of faster computation, Efficient Transformers demonstrate an impressive variety of approaches – models attaining sub-quadratic attention complexity can utilize a notion of sparsity or a low-rank approximation of inputs to reduce the number of attended keys; other ways to reduce complexity include locality-sensitive hashing, key pooling, additional memory to store information in compacted or hybridization with other architectures, such as CNN. Often based on a strong mathematical basis, kernelized approaches allow for the approximation of attention with linear complexity while retaining high accuracy. Therefore, in the present paper, we aim to expand the idea of trainable kernel methods to approximate the self-attention mechanism of the Transformer architecture.


page 1

page 2

page 3

page 4


Beyond Nyströmformer – Approximation of self-attention by Spectral Shifting

Transformer is a powerful tool for many natural language tasks which is ...

On The Computational Complexity of Self-Attention

Transformer architectures have led to remarkable progress in many state-...

Softmax-free Linear Transformers

Vision transformers (ViTs) have pushed the state-of-the-art for various ...

Combiner: Full Attention Transformer with Sparse Computation Cost

Transformers provide a class of expressive architectures that are extrem...

Linear Complexity Randomized Self-attention Mechanism

Recently, random feature attentions (RFAs) are proposed to approximate t...

Centroid Transformers: Learning to Abstract with Attention

Self-attention, as the key block of transformers, is a powerful mechanis...

Sparse Attention Acceleration with Synergistic In-Memory Pruning and On-Chip Recomputation

As its core computation, a self-attention mechanism gauges pairwise corr...