Summary of Partial-multivariate Model For Forecasting, by Jaehoon Lee et al.
Partial-Multivariate Model for Forecasting
by Jaehoon Lee, Hankook Lee, Sungik Choi, Sungjun Cho, Moontae Lee
First submitted to arxiv on: 19 Aug 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed paper introduces Partial-Multivariate (PM) models that leverage inter-feature information while avoiding the complexity of complete-multivariate approaches. The authors develop a Transformer-based PM model called PMformer, which captures partial relationships within subsets of all features. Experimental results show that PMformer outperforms univariate and complete-multivariate models on forecasting tasks, with a theoretical rationale and empirical analysis supporting its superiority. Additionally, an inference technique is proposed to further enhance forecasting accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a problem in time-series forecasting by introducing a new type of model called Partial-Multivariate (PM) models. The PM models are between two other types: univariate and complete-multivariate. Univariate models ignore information from multiple features, while complete-multivariate models try to use all the information. But sometimes, this second approach doesn’t work as well. The researchers created a new model called PMformer that is like a middle ground. They tested it and found that it does better than the other two types of models. This new model also has some nice features: it’s efficient and can still be used even if some of the information is missing. |
Keywords
» Artificial intelligence » Inference » Time series » Transformer
