2021-arxiv Gated Transformer Networks for Multivariate Time Series Classification

Motivation

In this work, researchers explored a simple extension of the current Transformer Networks with gating, named Gated Transformer Networks (GTN) for the multivariate time series classification problem.

Method

Actually, a two-towers transformers acquiring position & channel attention utilizing transposing time and position axis was proposed, and the pipeline was as follow:

Results


2021-arxiv Gated Transformer Networks for Multivariate Time Series Classification
https://firrice.github.io/posts/2024-06-26-6/
Author
firrice
Posted on
June 26, 2024
Licensed under