Summary of Units: a Unified Multi-task Time Series Model, by Shanghua Gao et al.
UniTS: A Unified Multi-Task Time Series Model
by Shanghua Gao, Teddy Koker, Owen Queen, Thomas Hartvigsen, Theodoros Tsiligkaridis, Marinka Zitnik
First submitted to arxiv on: 29 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 This abstract proposes UniTS, a unified multi-task time series model that combines predictive and generative tasks within a single framework. The model utilizes task tokenization to integrate these tasks and employs a modified transformer block to capture universal time series representations. This allows for transferability from a heterogeneous pre-training dataset to various downstream datasets with different task specifications and data domains. UniTS outperforms multiple state-of-the-art models, including adapted text-based LLMs, on 38 datasets across human activity sensors, healthcare, engineering, and finance. It also demonstrates strong few-shot and prompt capabilities when applied to new domains and tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary UniTS is a special kind of computer program that can help predict or generate time series data (like stock prices or weather patterns). Right now, there are many different programs that do this, but they’re all good at one thing and not very good at anything else. UniTS tries to change that by being able to do multiple things well, like forecasting, classification, anomaly detection, and imputation. It does this by using a special way of representing time series data called task tokenization. The program was tested on many different datasets and performed better than other similar programs. |
Keywords
* Artificial intelligence * Anomaly detection * Classification * Few shot * Multi task * Prompt * Time series * Tokenization * Transferability * Transformer
