Learning from Summarized Data: Gaussian Process Regression with Sample Quasi-Likelihood
Authors: Yuta Shikuri
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Theoretical and experimental results demonstrate that the approximation performance is influenced by the granularity of summarized data relative to the length scale of covariance functions. Experiments on a real-world dataset highlight the practicality of our method for spatial modeling. |
| Researcher Affiliation | Industry | Yuta Shikuri Tokio Marine Holdings, Inc. Tokyo, Japan EMAIL |
| Pseudocode | No | The paper describes mathematical formulations and procedures but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks or figures. |
| Open Source Code | No | The text states, 'The code was implemented in the python programming language 3.7.3 version.' However, it does not provide an explicit statement of code release or a link to a repository for the methodology described in the paper. It refers to third-party tools like SciPy and GPy but not to the authors' own implementation code. |
| Open Datasets | Yes | We investigate the usage of our method in spatial modeling tasks, using the California housing dataset 1. 1https://lib.stat.cmu.edu/datasets/ |
| Dataset Splits | Yes | The dataset was randomly shuffled and then split to obtain the n training data points and n = 20640 n test data points. [...] Let n = 10000. |
| Hardware Specification | Yes | We used a 64bit Windows machine with the Intel Core i9-9900K @ 3.60 GHz and 64 GB of RAM. |
| Software Dependencies | Yes | This optimization was implemented with version 1.7.3 of the Sci Py software 2, with the parameters set to their default values. We used a 64bit Windows machine with the Intel Core i9-9900K @ 3.60 GHz and 64 GB of RAM. The code was implemented in the python programming language 3.7.3 version. As a hyperparameter, we also optimized the variance σ (0, ) of the Gaussian likelihood. [...] implemented in version 1.12.0 of the GPy software 3. |
| Experiment Setup | Yes | The mean function was τ( ) = g 1( 1 n Pn i=1 yi). The covariance functions were designed by multiplying a constant kernel with the functions used in fig. 3 and adding white noise. [...] The hyperparameters of the covariance functions were optimized using the L-BFGS-B algorithm, starting from an initial value of 1. This optimization was implemented with version 1.7.3 of the Sci Py software 2, with the parameters set to their default values. [...] As a hyperparameter, we also optimized the variance σ (0, ) of the Gaussian likelihood. [...] The domain [32.54, 41.95] [ 124.35, 114.31] in latitude and longitude was divided into girds. The data points within each grid were summarized, with the centers of the grids serving as the representative features, and the summary statistics being the sample means. |