Summary of Negative-binomial Randomized Gamma Markov Processes For Heterogeneous Overdispersed Count Time Series, by Rui Huang et al.
Negative-Binomial Randomized Gamma Markov Processes for Heterogeneous Overdispersed Count Time Series
by Rui Huang, Sikun Yang, Heinz Koeppl
First submitted to arxiv on: 29 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
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 research addresses the challenge of modeling count-valued time series, which is crucial in various physical and social domains. The authors introduce Poisson gamma dynamical systems (PGDSs) that excel in capturing latent transition structures and bursty dynamics behind count sequences. Compared to linear dynamical system-based methods, PGDSs demonstrate superior performance in data imputation and prediction. However, the limitations of PGDSs, including its inability to capture heterogeneous overdispersed behaviors, are addressed by proposing a negative-binomial-randomized gamma Markov process that improves predictive performance and inference algorithm convergence. The authors also develop methods for estimating factor-structured and graph-structured transition dynamics, allowing for more explainable latent structures compared to PGDSs. Overall, the proposed method shows superior performance in imputing missing data and forecasting future observations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about finding new ways to analyze time series data that includes counts of events. Right now, there are limited methods available that can capture complex patterns in this type of data. The researchers propose two new approaches: Poisson gamma dynamical systems (PGDSs) and negative-binomial-randomized gamma Markov processes. PGDSs are good at capturing sudden changes or bursts in the data, but they have limitations when it comes to understanding why these changes happen. By combining PGDSs with other techniques, the researchers can get a better understanding of what’s going on in the data and make more accurate predictions. |
Keywords
* Artificial intelligence * Inference * Time series
