Transformer Multivariate Forecasting: Less is More?
Contents
- Abstract
- Introduction
- PCA+Transformer in TSF
0. Abstract
Inherent complexity of TS datasets
\(\rightarrow\) This paper: focuses on reducing redundant information
- to elevate forecasting accuracy
- while optimizing runtime efficiency.
Propose Transformer + Principal Component Analysis (PCA)
1. Introduction
Task : LTSF with MTS
Trend of Google Scholar’s publications on Transformer techniques since 2017.
Dimension reduction
- Existing approaches to dimensionality reduction in datasets for transformer models remain insufficient.
Contribution
- Present aa benchmark test on transformer-based MTS forecasting with PCA
- Proposed framework can significantly enhance LTSF performance
- Comprehensive study
- Detailed and interpretable insights for model transparency,
- Interpretability and explainability of multivariate analysis
2. PCA+Transformer in TSF
