Multi-resolution Multi-task Gaussian Processes
Authors: Oliver Hamelijnck, Theodoros Damoulas, Kangrui Wang, Mark Girolami
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the competitiveness of MRGPs on synthetic settings and on the challenging problem of hyper-local estimation of air pollution levels across London from multiple sensing modalities operating at disparate spatio-temporal resolutions. |
| Researcher Affiliation | Academia | Oliver Hamelijnck The Alan Turing Institute Department of Computer Science University of Warwick ohamelijnck@turing.ac.uk Theodoros Damoulas The Alan Turing Institute Depts. of Computer Science & Statistics University of Warwick tdamoulas@turing.ac.uk Kangrui Wang The Alan Turing Institute Department of Statistics University of Warwick kwang@turing.ac.uk Mark A. Girolami The Alan Turing Institute Department of Engineering University of Cambridge mgirolami@turing.ac.uk |
| Pseudocode | Yes | Algorithm 1 Inference of MR-GPRN [...] Algorithm 2 Inference procedure for MR-DGP |
| Open Source Code | Yes | Further analysis is provided in the Appendix and code is available at https: //github.com/ohamelijnck/multi_res_gps. [...] Codebase and datasets to reproduce results are available at https://github.com/ohamelijnck/multi_ res_gps |
| Open Datasets | Yes | observations coming from ground point sensors from the London Air Quality Network (LAQN). These sensors provide hourly readings of NO2. Secondly we use observations arising from a global satellite model [17] that provide hourly data at a spatial resolution of 7km 7km and provide 48 hour forecasts. [...] Codebase and datasets to reproduce results are available at https://github.com/ohamelijnck/multi_ res_gps |
| Dataset Splits | No | The paper mentions removing a 2-day region for testing and uses temporal ranges for training and prediction, but it does not specify explicit training/validation/test split percentages, sample counts, or a detailed splitting methodology for reproducibility. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU/CPU models, memory, or cloud instance types used for running the experiments. |
| Software Dependencies | No | The paper states 'Experiments are coded in Tensor Flow' but does not provide specific version numbers for TensorFlow or any other software dependencies. |
| Experiment Setup | No | The paper mentions optimizing variational and hyperparameters for training and a layer-by-layer optimization strategy for MR-DGP, but it does not provide specific numerical values for hyperparameters (e.g., learning rate, batch size, number of epochs) or detailed system-level training settings in the main text. |