Summary of Towards Foundation Time Series Model: to Synthesize or Not to Synthesize?, by Kseniia Kuvshinova et al.
Towards Foundation Time Series Model: To Synthesize Or Not To Synthesize?
by Kseniia Kuvshinova, Olga Tsymboi, Alina Kostromina, Dmitry Simakov, Elizaveta Kovtun
First submitted to arxiv on: 4 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- 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 In this paper, researchers tackle the problem of forecasting large amounts of time series data without retraining separate models for each one. They propose establishing a foundation model that can work in zero-shot and few-shot regimes, but first, they need to determine what training dataset is best suited for such a model. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The researchers are trying to solve a common problem in time series modeling where you have many datasets to forecast, but it’s not feasible to train separate models for each one. They want to create a foundation model that can work without needing to be trained on every individual dataset. But before they can do that, they need to figure out what kind of data should be used to train this foundation model. |
Keywords
* Artificial intelligence * Few shot * Time series * Zero shot
