Poisson-Randomized Gamma Dynamical Systems
Authors: Aaron Schein, Scott Linderman, Mingyuan Zhou, David Blei, Hanna Wallach
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We study the inductive bias of the PRGDS by comparing its smoothing and forecasting performance to that of the PGDS and two other baselines on a range of real-world count data sets of text, international events, and neural spike data. For smoothing, we find that the PRGDS performs better than or similarly to the PGDS; for forecasting, we find the converse relationship. Both models outperform the other baselines. |
| Researcher Affiliation | Collaboration | Aaron Schein Data Science Institute Columbia University Scott W. Linderman Department of Statistics Stanford University Mingyuan Zhou Mc Combs School of Business University of Texas at Austin David M. Blei Department of Statistics Columbia University Hanna Wallach Microsoft Research New York, NY |
| Pseudocode | No | The paper describes mathematical models and inference steps, but it does not include a clearly labeled pseudocode or algorithm block. |
| Open Source Code | Yes | The Bessel distribution can be sampled efficiently [53]; our Cython implementation is available online.1 https://github.com/aschein/PRGDS |
| Open Datasets | Yes | The matrices in these studies were based on three text data sets Neur IPS papers [57], DBLP abstracts [58], and State of the Union (SOTU) speeches [59] where y(t) v is the number of times word v occurs in time step t, and two international event data sets GDELT [60] and ICEWS [61] where y(t) v is the number of times sender receiver pair v interacted during time step t. |
| Dataset Splits | Yes | For each data set Y (1), . . . , Y (T ), the counts Y (t) in randomly selected time steps are held out. Additionally, the counts in the last two time steps are always held out. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU/GPU models, memory, or cloud instance types) used for running its experiments. |
| Software Dependencies | No | The paper mentions a 'Cython implementation' but does not provide specific version numbers for Cython or any other software libraries or dependencies. |
| Experiment Setup | Yes | We fit each model to each data set and mask using two independent chains of 4,000 MCMC iterations, saving every 50th posterior sample after the first 1,000 iterations to compute the information rate. ... Following Schein et al. [22], we set K =100 for all models. |