--- title: "Yee Whye Teh — author profile and papers" description: "Yee Whye Teh, Google: 2 papers on Elamur Research (2006–2016), h-index 53 — affiliations, topics, co-authors and who cites them in our library." canonical: "https://research.elamur.ai/free/en/authors/yee-whye-teh/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/yee-whye-teh/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Yee Whye Teh # Yee Whye Teh Yee Whye Teh works on probabilistic machine learning. The papers here cover Bayesian modelling of language and methods for training models that contain random discrete choices. A recurring concern is giving practical techniques a principled probabilistic footing. One paper explains a well-performing smoothing heuristic as inference in a generative model. Another replaces discrete variables with continuous stand-ins so that gradients can be computed. The 2006 paper, [A Hierarchical Bayesian Language Model Based On Pitman-Yor Processes](https://md.research.elamur.ai/free/en/papers/teh-2006-hierarchical-bayesian-language-model-based/), revisits n-gram smoothing, an area of [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/). For every context, the model places a Pitman-Yor prior over the following word, centred on the distribution for that context with its oldest word removed. Contexts then form a suffix tree that shares statistical strength between related histories. The prior was chosen to reproduce the power-law growth of vocabularies, and the model accounts for the strong performance of Kneser-Ney smoothing. The newer of the two, [The Concrete Distribution](https://md.research.elamur.ai/free/en/papers/maddison-2016-concrete-distribution-continuous-relaxation-discrete/) from 2016, was written with Maddison and Mnih and turns to stochastic computation graphs. The reparameterization trick does not work for discrete nodes, so such models usually fall back on score-function gradient estimators, which tend to be noisy. The paper relaxes categorical variables into continuous ones, defining distributions over the simplex whose densities can be written in closed form. With 1,569 citations, it is the most cited of Teh's papers here. Teh is at the [University of Oxford](https://md.research.elamur.ai/free/en/universities/university-of-oxford/) and has current ties to [Google](https://md.research.elamur.ai/free/en/companies/google/) and the [National University of Singapore](https://md.research.elamur.ai/free/en/universities/national-university-of-singapore/). Teh has also published with [Google DeepMind](https://md.research.elamur.ai/free/en/companies/google-deepmind/) in London. Earlier papers come from [University College London](https://md.research.elamur.ai/free/en/universities/university-college-london/) and the Gatsby Computational Neuroscience Unit, and from the [University of Toronto](https://md.research.elamur.ai/free/en/universities/university-of-toronto/), where Teh studied. Yee Whye Teh has 2 papers in our library, in 2006–2016, most of them on [Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/); often with [Chris J. Maddison](https://md.research.elamur.ai/free/en/authors/chris-j-maddison/). 2 papers in our library · 2006–2016 current affiliation [Google](https://md.research.elamur.ai/free/en/companies/google/) · Mountain View United States; [National University of Singapore](https://md.research.elamur.ai/free/en/universities/national-university-of-singapore/) · Singapore Singapore; [University of Oxford](https://md.research.elamur.ai/free/en/universities/university-of-oxford/) · Oxford United Kingdom [OpenAlex](https://openalex.org/A5064373793) · [ORCID](https://orcid.org/0000-0001-5365-6933) · [Wikidata](https://www.wikidata.org/wiki/Q44585452) · [**h-index 53**](https://openalex.org/authors/A5064373793) · [317 works](https://openalex.org/works?filter=author.id:A5064373793) · [17,545 citations](https://openalex.org/works?filter=author.id:A5064373793&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [2 works in our library](https://md.research.elamur.ai/free/en/authors/yee-whye-teh/#papers) · [688 citations in our library](https://md.research.elamur.ai/free/en/authors/yee-whye-teh/cited-by/) · 4 more works in our library · [⭐ Paper Star rank #1126](https://md.research.elamur.ai/free/en/paper-stars/page/12/#star-1126) The profile is the library's match of the name; two people with one name are not always told apart. Elsewhere: [Google Scholar](https://scholar.google.com/citations?view_op=search_authors&mauthors=Yee+Whye+Teh) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Yee+Whye+Teh) · [dblp](https://dblp.org/search?q=Yee+Whye+Teh) · [ACL Anthology](https://aclanthology.org/search/?q=Yee+Whye+Teh) ## Affiliations [🎓**University of Oxford** Oxford · United Kingdom 2012–2026 · 134 papers*current**employment from 2012 (Wikidata)*](https://md.research.elamur.ai/free/en/universities/university-of-oxford/) · [🏢**Google** Mountain View · United States 2016–2026 · 6 papers*current**employment from 2016 (Wikidata)*](https://md.research.elamur.ai/free/en/companies/google/) · [🎓**National University of Singapore** Singapore · Singapore 2005–2026 · 4 papers*current**employment 2005–2006 (Wikidata)*](https://md.research.elamur.ai/free/en/universities/national-university-of-singapore/) · [🎓**University College London** London · United Kingdom 2007–2020 · 45 papers*employment 2007–2012 (Wikidata)*](https://md.research.elamur.ai/free/en/universities/university-college-london/) · [🔬**Gatsby Computational Neuroscience Unit** London · United Kingdom 2007–2020 · 26 papers](https://ror.org/0269epf40) · [🏢**Google DeepMind** London · United Kingdom 2017–2025 · 21 papers](https://md.research.elamur.ai/free/en/companies/google-deepmind/) · [🎓**University of Toronto** Toronto · Canada 1998–2015 · 9 papers*education (Wikidata)*](https://md.research.elamur.ai/free/en/universities/university-of-toronto/) · [🎓**University of California, Berkeley** Berkeley · United States 2003–2012 · 8 papers*employment 2003–2004 (Wikidata)*](https://md.research.elamur.ai/free/en/universities/university-of-california-berkeley/) · [🎓**University of Turin** Turin · Italy 2014–2015 · 2 papers](https://md.research.elamur.ai/free/en/universities/university-of-turin/) · [🎓**University of Waterloo** Waterloo · Canada *education (Wikidata)*](https://md.research.elamur.ai/free/en/universities/university-of-waterloo/) From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 7 more institutions on single bylines are not listed. ## Topics [Language models & pretraining 1](https://md.research.elamur.ai/free/en/topics/language-models/) ## Works by year ### 2010s [**2016** 1 paper](https://md.research.elamur.ai/free/en/papers/maddison-2016-concrete-distribution-continuous-relaxation-discrete/) ### 2000s [**2006** 1 paper](https://md.research.elamur.ai/free/en/papers/teh-2006-hierarchical-bayesian-language-model-based/) ## Papers The most cited first (OpenAlex and our library). 1. **[The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables](https://md.research.elamur.ai/free/en/papers/maddison-2016-concrete-distribution-continuous-relaxation-discrete/)** *(2016 · 1,569 citations)* The reparameterization trick lets gradients pass through continuous random nodes in stochastic computation graphs. 2. **[A Hierarchical Bayesian Language Model Based On Pitman-Yor Processes](https://md.research.elamur.ai/free/en/papers/teh-2006-hierarchical-bayesian-language-model-based/)** *(2006 · 571 citations)* The paper addresses a long-standing gap in n-gram language modelling. The strongest smoothing techniques, such as Kneser-Ney variants, are justified mostly by empirical results. Earlier Bayesian models, which have explicit prior assumptions, performed poorly. Teh proposes the hierarchical Pitman-Yor language model… ## Also in our library 4 more works of theirs are in our library without a page here yet; they will appear as we publish them. ## Co-authors here 1 co-author with a page, the most papers together first. - [Chris J. Maddison](https://md.research.elamur.ai/free/en/authors/chris-j-maddison/) ## Cited by Papers of our library cite their works 688 times (of the 140,884 papers whose bibliographies we read); [28 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/yee-whye-teh/cited-by/).