{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/a-unifying-framework-for-gaussian-process","title":"A Unifying Framework for Gaussian Process Pseudo-Point Approximations using Power Expectation Propagation","arxiv_id":"1605.07066","date":"2016-05-23","proceeding":null,"authors":["Thang D. Bui","Josiah Yan","Richard E. Turner"],"abstract":"Gaussian processes (GPs) are flexible distributions over functions that\nenable high-level assumptions about unknown functions to be encoded in a\nparsimonious, flexible and general way. Although elegant, the application of\nGPs is limited by computational and analytical intractabilities that arise when\ndata are sufficiently numerous or when employing non-Gaussian models.\nConsequently, a wealth of GP approximation schemes have been developed over the\nlast 15 years to address these key limitations. Many of these schemes employ a\nsmall set of pseudo data points to summarise the actual data. In this paper, we\ndevelop a new pseudo-point approximation framework using Power Expectation\nPropagation (Power EP) that unifies a large number of these pseudo-point\napproximations. Unlike much of the previous venerable work in this area, the\nnew framework is built on standard methods for approximate inference\n(variational free-energy, EP and Power EP methods) rather than employing\napproximations to the probabilistic generative model itself. In this way, all\nof approximation is performed at `inference time' rather than at `modelling\ntime' resolving awkward philosophical and empirical questions that trouble\nprevious approaches. Crucially, we demonstrate that the new framework includes\nnew pseudo-point approximation methods that outperform current approaches on\nregression and classification tasks.","url_abs":"http://arxiv.org/abs/1605.07066v3","url_pdf":"http://arxiv.org/pdf/1605.07066v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"a-unifying-framework-for-gaussian-process","repo_url":"https://github.com/thangbui/sparseGP_powerEP","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"gaussian-processes","task_name":"Gaussian Processes"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}