{"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/variational-zero-inflated-gaussian-processes","title":"Variational zero-inflated Gaussian processes with sparse kernels","arxiv_id":"1803.05036","date":"2018-03-13","proceeding":null,"authors":["Pashupati Hegde","Markus Heinonen","Samuel Kaski"],"abstract":"Zero-inflated datasets, which have an excess of zero outputs, are commonly\nencountered in problems such as climate or rare event modelling. Conventional\nmachine learning approaches tend to overestimate the non-zeros leading to poor\nperformance. We propose a novel model family of zero-inflated Gaussian\nprocesses (ZiGP) for such zero-inflated datasets, produced by sparse kernels\nthrough learning a latent probit Gaussian process that can zero out kernel rows\nand columns whenever the signal is absent. The ZiGPs are particularly useful\nfor making the powerful Gaussian process networks more interpretable. We\nintroduce sparse GP networks where variable-order latent modelling is achieved\nthrough sparse mixing signals. We derive the non-trivial stochastic variational\ninference tractably for scalable learning of the sparse kernels in both models.\nThe novel output-sparse approach improves both prediction of zero-inflated data\nand interpretability of latent mixing models.","url_abs":"http://arxiv.org/abs/1803.05036v1","url_pdf":"http://arxiv.org/pdf/1803.05036v1.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":"variational-zero-inflated-gaussian-processes","repo_url":"https://github.com/hegdepashupati/zero-inflated-gp","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"gaussian-processes","task_name":"Gaussian Processes"},{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[{"method_slug":"gaussian-process","method_name":"Gaussian Process"},{"method_slug":"interpretability","method_name":"Interpretability"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}