Steven D. Barnett, Lauren J. Beesley, Annie S. Booth, Robert B. Gramacy, Dave Osthus
{"title":"加性和深高斯过程的单调翘曲","authors":"Steven D. Barnett, Lauren J. Beesley, Annie S. Booth, Robert B. Gramacy, Dave Osthus","doi":"arxiv-2408.01540","DOIUrl":null,"url":null,"abstract":"Gaussian processes (GPs) are canonical as surrogates for computer experiments\nbecause they enjoy a degree of analytic tractability. But that breaks when the\nresponse surface is constrained, say to be monotonic. Here, we provide a\nmono-GP construction for a single input that is highly efficient even though\nthe calculations are non-analytic. Key ingredients include transformation of a\nreference process and elliptical slice sampling. We then show how mono-GP may\nbe deployed effectively in two ways. One is additive, extending monotonicity to\nmore inputs; the other is as a prior on injective latent warping variables in a\ndeep Gaussian process for (non-monotonic, multi-input) non-stationary surrogate\nmodeling. We provide illustrative and benchmarking examples throughout, showing\nthat our methods yield improved performance over the state-of-the-art on\nexamples from those two classes of problems.","PeriodicalId":501215,"journal":{"name":"arXiv - STAT - Computation","volume":"58 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Monotonic warpings for additive and deep Gaussian processes\",\"authors\":\"Steven D. Barnett, Lauren J. Beesley, Annie S. Booth, Robert B. Gramacy, Dave Osthus\",\"doi\":\"arxiv-2408.01540\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Gaussian processes (GPs) are canonical as surrogates for computer experiments\\nbecause they enjoy a degree of analytic tractability. But that breaks when the\\nresponse surface is constrained, say to be monotonic. Here, we provide a\\nmono-GP construction for a single input that is highly efficient even though\\nthe calculations are non-analytic. Key ingredients include transformation of a\\nreference process and elliptical slice sampling. We then show how mono-GP may\\nbe deployed effectively in two ways. One is additive, extending monotonicity to\\nmore inputs; the other is as a prior on injective latent warping variables in a\\ndeep Gaussian process for (non-monotonic, multi-input) non-stationary surrogate\\nmodeling. We provide illustrative and benchmarking examples throughout, showing\\nthat our methods yield improved performance over the state-of-the-art on\\nexamples from those two classes of problems.\",\"PeriodicalId\":501215,\"journal\":{\"name\":\"arXiv - STAT - Computation\",\"volume\":\"58 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-02\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - STAT - Computation\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2408.01540\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - STAT - Computation","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2408.01540","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
高斯过程(GPs)是计算机实验的典型代表,因为它们具有一定程度的可分析性。但是,当响应面受到约束,例如必须是单调的时候,这种可分析性就不复存在了。在这里,我们为单一输入提供了一种单 GP 结构,即使计算是非解析的,它也非常高效。其关键要素包括推理过程的转换和椭圆切片采样。然后,我们展示了如何以两种方式有效地部署单GP。一种是加法,将单调性扩展到更多输入;另一种是作为深高斯过程中注入式潜翘变量的先验,用于(非单调、多输入)非稳态代理建模。我们通篇提供了说明性和基准示例,表明我们的方法在这两类问题的示例中取得了优于最先进方法的性能。
Monotonic warpings for additive and deep Gaussian processes
Gaussian processes (GPs) are canonical as surrogates for computer experiments
because they enjoy a degree of analytic tractability. But that breaks when the
response surface is constrained, say to be monotonic. Here, we provide a
mono-GP construction for a single input that is highly efficient even though
the calculations are non-analytic. Key ingredients include transformation of a
reference process and elliptical slice sampling. We then show how mono-GP may
be deployed effectively in two ways. One is additive, extending monotonicity to
more inputs; the other is as a prior on injective latent warping variables in a
deep Gaussian process for (non-monotonic, multi-input) non-stationary surrogate
modeling. We provide illustrative and benchmarking examples throughout, showing
that our methods yield improved performance over the state-of-the-art on
examples from those two classes of problems.