Continual Low-Rank Scaled Dot-product Attention

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Picón, Ginés Carreto, Oleksiienko, Illia, Hedegaard, Lukas, Bakhtiarnia, Arian, Iosifidis, Alexandros
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918105103466496
author Picón, Ginés Carreto
Oleksiienko, Illia
Hedegaard, Lukas
Bakhtiarnia, Arian
Iosifidis, Alexandros
author_facet Picón, Ginés Carreto
Oleksiienko, Illia
Hedegaard, Lukas
Bakhtiarnia, Arian
Iosifidis, Alexandros
contents Transformers are widely used for their ability to capture data relations in sequence processing, with great success for a wide range of static tasks. However, the computational and memory footprint of their main component, i.e., the Scaled Dot-product Attention, is commonly overlooked. This makes their adoption in applications involving stream data processing with constraints in response latency, computational and memory resources infeasible. Some works have proposed methods to lower the computational cost of Transformers, i.e. low-rank approximations, sparsity in attention, and efficient formulations for Continual Inference. In this paper, we introduce a new formulation of the Scaled Dot-product Attention based on the Nyström approximation that is suitable for Continual Inference. In experiments on Online Audio Classification and Online Action Detection tasks, the proposed Continual Scaled Dot-product Attention can lower the number of operations by up to three orders of magnitude compared to the original Transformers while retaining the predictive performance of competing models.
format Preprint
id arxiv_https___arxiv_org_abs_2412_03214
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Continual Low-Rank Scaled Dot-product Attention
Picón, Ginés Carreto
Oleksiienko, Illia
Hedegaard, Lukas
Bakhtiarnia, Arian
Iosifidis, Alexandros
Computer Vision and Pattern Recognition
Machine Learning
Transformers are widely used for their ability to capture data relations in sequence processing, with great success for a wide range of static tasks. However, the computational and memory footprint of their main component, i.e., the Scaled Dot-product Attention, is commonly overlooked. This makes their adoption in applications involving stream data processing with constraints in response latency, computational and memory resources infeasible. Some works have proposed methods to lower the computational cost of Transformers, i.e. low-rank approximations, sparsity in attention, and efficient formulations for Continual Inference. In this paper, we introduce a new formulation of the Scaled Dot-product Attention based on the Nyström approximation that is suitable for Continual Inference. In experiments on Online Audio Classification and Online Action Detection tasks, the proposed Continual Scaled Dot-product Attention can lower the number of operations by up to three orders of magnitude compared to the original Transformers while retaining the predictive performance of competing models.
title Continual Low-Rank Scaled Dot-product Attention
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2412.03214