Summary of Better Batch For Deep Probabilistic Time Series Forecasting, by Vincent Zhihao Zheng et al.
Better Batch for Deep Probabilistic Time Series Forecasting
by Vincent Zhihao Zheng, Seongjin Choi, Lijun Sun
First submitted to arxiv on: 26 May 2023
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG)
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 paper proposes an innovative training method for deep probabilistic time series forecasting that incorporates error autocorrelation to enhance accuracy and uncertainty quantification. The existing models often oversimplify the problem by assuming a time-independent error process and overlooking serial correlation. To address this limitation, the authors construct a mini-batch as a collection of consecutive time series segments for model training, which learns a time-varying covariance matrix over each mini-batch to encode error correlation among adjacent time steps. This approach is evaluated on two different neural forecasting models and multiple public datasets, showing notable improvements in predictive accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper develops a new way to predict future events based on past data. It wants to improve the accuracy of these predictions by considering how errors are connected over time. The authors train their model using small groups of consecutive data points, which helps it learn how errors change over time. This approach is tested on different models and datasets, showing that it can lead to more accurate predictions. |
Keywords
* Artificial intelligence * Time series
