Gaussian Approximation of Collective Graphical Models
Authors: Liping Liu, Daniel Sheldon, Thomas Dietterich
ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we evaluate the performance of our method and compare it to the MAP approximation of Sheldon, Sun, Kumar, and Dietterich (2013). The evaluation data are generated from the bird migration model introduced in Sheldon et al. (2013). This model simulates the migration of a population of M birds on an L = ℓ ℓmap. |
| Researcher Affiliation | Academia | Li-Ping Liu1 LIULI@EECS.OREGONSTATE.EDU Daniel Sheldon2 SHELDON@CS.UMASS.EDU Thomas G. Dietterich1 TGD@EECS.OREGONSTATE.EDU 1School of EECS, Oregon State University, Corvallis, OR 97331 USA 2University of Massachusetts, Amherst, MA 01002 and Mount Holyoke College, South Hadley, MA 01075 |
| Pseudocode | No | No explicit pseudocode or algorithm blocks were found. |
| Open Source Code | No | No statement regarding open-source code availability or links to code repositories were found. |
| Open Datasets | No | The evaluation data are generated from the bird migration model introduced in Sheldon et al. (2013). No link or specific access details are provided for this simulated data. |
| Dataset Splits | No | No specific dataset splits (training, validation, test) or cross-validation setup details were provided. |
| Hardware Specification | No | No specific hardware details (e.g., GPU/CPU models, memory specifications) used for experiments were provided. |
| Software Dependencies | No | No specific software dependencies with version numbers were mentioned. |
| Experiment Setup | Yes | In this table, we fixed L = 36, set the logistic regression coefficient vector w = (1, 2, 2, 2), and varied the population size N {36, 360, 1080, 3600}. At each time step t, the data generation model generates an observation vector yt of length L which contains noisy counts of birds at all map cells at time t, nt. The observed counts are generated by a Poisson distribution with unit intensity. We estimate the expected values by running the MCMC method (Sheldon & Dietterich, 2011) for a burn-in period of 1 million Gibbs iterations and then collecting samples from 10 million Gibbs iterations and averaging the results. |