Bayesian classification: methodology, algorithms and applications Subhashis Ghoshal will visit in July 2025 and present his work on Bayesian semi-supervised learning. The event will also feature short talks from the Schools of Maths and Informatics.SpeakersCecilia Balocchi (School of Mathematics): Bayesian image segmentation of remote sensing imagesHeng Guo (School of Informatics): Counting and sampling from a computational complexity perspectiveProf Subhashis Ghoshal (North Carolina State University): Bayesian Semi-supervised Multicategory Classification under Nonparanormality Registration Please register your interest to attend for catering purposes. Register Now! Schedule 9:15: Registration9:25: Welcome9:30-10:00: Cecilia Balocchi (School of Mathematics, UoE) Title: Bayesian image segmentation of remote sensing images10:00-10:30: Heng Guo (School of Informatics, UoE) Title: Counting and sampling from a computational complexity perspective10:30-11:00: Coffee break11:00-12:00: Subhashis Ghoshal (University of North Carolina) Title: Bayesian Semi-supervised Multicategory Classification under NonparanormalityAbstract: Semi-supervised learning is a machine learning technique combining supervised and unsupervised learning by using labeled and unlabeled data to train statistical models for classification and regression tasks. This paper addresses the problem of semi-supervised binary classification, assuming that the underlying data has only a few observations labeled in each class. Some methods are developed for semi-supervised classification outside the Bayesian domain, but most works in the Bayesian domain use Gaussian mixture models. However, the assumption that the subpopulations are Gaussian may not be realistic in some situations. We generalize the data-generating process to the nonparanormality setting: the observations result from an unknown component-wise monotone increasing transformation applied to a hidden layer of multivariate normal latent variables. We put a prior distribution on the transformation functions through B-splines that naturally maintains monotonicity and satisfies required identifiability constraints. We use a Gibbs sampler to coordinate draws from the posterior distribution of four objects: the missing labels, the coefficients of the B-spline expansions of the transformation functions, the parameters of the multivariate normal distributions of the component populations, and the population mixing proportions. The posterior draws of these objects use the Bayes formula for categories, Hamiltonian Monte Carlo, normal-normal conjugacy, and beta-binomial conjugacy, respectively. Using a low-density at separation assumption, we tune the number of terms in the B-spline expansions. We evaluate the performance of the proposed method based on extensive simulated data. We conclude that the proposed method gives low classification error rates even when the nonparanormality assumption is violated and outperforms many state-of-the-art semi-supervised machine learning techniques. The method performs well on several benchmark binary classification datasets. Professor SUBHASHIS GHOSHAL Subhashis Ghoshal is a Goodnight Distinguished Professor of Statistics at North Carolina State University, Raleigh. His research interests span many areas including Bayesian statistics, asymptotics, nonparametrics and high dimensional models, with diverse applications. In particular, his pioneering work on concentration of posterior distributions led to theoretical understanding of nonparametric Bayesian procedures. He was honoured with fellowship from the Institute of Mathematical Statistics (2006), American Statistical Association (2010) and International Society for Bayesian Analysis (2016). He has received several awards, including the IMS Medallion lecture (2017), NSF Career Award (2003), Sigma-Xi Research Award (2004), International Indian Statistical Association Young Researcher Award (2007) and Cavell Brownie Mentoring Award (2015). He was awarded De Groot Prize for the best book on Statistical Science in 2019 given by the International Society for Bayesian Analysis for the book “Fundamentals of Nonparametric Bayesian Inference” (co-author: Aad van der Vaart), published by Cambridge University Press, 2017. In July-August 2025 Professor SUBHASHIS Ghoshal will be visiting the UK as a Distinguished Rothschild Visiting Fellow at the Isaac Newton Institute, University of Cambridge. Webpage: Subhashis Ghoshal Jul 21 2025 09.00 - 12.30 Bayesian classification: methodology, algorithms and applications Subhashis Ghoshal will visit in July 2025 and present his work on Bayesian semi-supervised learning. The event will also feature short talks from the Schools of Maths and Informatics. Bayes Centre G.03
Bayesian classification: methodology, algorithms and applications Subhashis Ghoshal will visit in July 2025 and present his work on Bayesian semi-supervised learning. The event will also feature short talks from the Schools of Maths and Informatics.SpeakersCecilia Balocchi (School of Mathematics): Bayesian image segmentation of remote sensing imagesHeng Guo (School of Informatics): Counting and sampling from a computational complexity perspectiveProf Subhashis Ghoshal (North Carolina State University): Bayesian Semi-supervised Multicategory Classification under Nonparanormality Registration Please register your interest to attend for catering purposes. Register Now! Schedule 9:15: Registration9:25: Welcome9:30-10:00: Cecilia Balocchi (School of Mathematics, UoE) Title: Bayesian image segmentation of remote sensing images10:00-10:30: Heng Guo (School of Informatics, UoE) Title: Counting and sampling from a computational complexity perspective10:30-11:00: Coffee break11:00-12:00: Subhashis Ghoshal (University of North Carolina) Title: Bayesian Semi-supervised Multicategory Classification under NonparanormalityAbstract: Semi-supervised learning is a machine learning technique combining supervised and unsupervised learning by using labeled and unlabeled data to train statistical models for classification and regression tasks. This paper addresses the problem of semi-supervised binary classification, assuming that the underlying data has only a few observations labeled in each class. Some methods are developed for semi-supervised classification outside the Bayesian domain, but most works in the Bayesian domain use Gaussian mixture models. However, the assumption that the subpopulations are Gaussian may not be realistic in some situations. We generalize the data-generating process to the nonparanormality setting: the observations result from an unknown component-wise monotone increasing transformation applied to a hidden layer of multivariate normal latent variables. We put a prior distribution on the transformation functions through B-splines that naturally maintains monotonicity and satisfies required identifiability constraints. We use a Gibbs sampler to coordinate draws from the posterior distribution of four objects: the missing labels, the coefficients of the B-spline expansions of the transformation functions, the parameters of the multivariate normal distributions of the component populations, and the population mixing proportions. The posterior draws of these objects use the Bayes formula for categories, Hamiltonian Monte Carlo, normal-normal conjugacy, and beta-binomial conjugacy, respectively. Using a low-density at separation assumption, we tune the number of terms in the B-spline expansions. We evaluate the performance of the proposed method based on extensive simulated data. We conclude that the proposed method gives low classification error rates even when the nonparanormality assumption is violated and outperforms many state-of-the-art semi-supervised machine learning techniques. The method performs well on several benchmark binary classification datasets. Professor SUBHASHIS GHOSHAL Subhashis Ghoshal is a Goodnight Distinguished Professor of Statistics at North Carolina State University, Raleigh. His research interests span many areas including Bayesian statistics, asymptotics, nonparametrics and high dimensional models, with diverse applications. In particular, his pioneering work on concentration of posterior distributions led to theoretical understanding of nonparametric Bayesian procedures. He was honoured with fellowship from the Institute of Mathematical Statistics (2006), American Statistical Association (2010) and International Society for Bayesian Analysis (2016). He has received several awards, including the IMS Medallion lecture (2017), NSF Career Award (2003), Sigma-Xi Research Award (2004), International Indian Statistical Association Young Researcher Award (2007) and Cavell Brownie Mentoring Award (2015). He was awarded De Groot Prize for the best book on Statistical Science in 2019 given by the International Society for Bayesian Analysis for the book “Fundamentals of Nonparametric Bayesian Inference” (co-author: Aad van der Vaart), published by Cambridge University Press, 2017. In July-August 2025 Professor SUBHASHIS Ghoshal will be visiting the UK as a Distinguished Rothschild Visiting Fellow at the Isaac Newton Institute, University of Cambridge. Webpage: Subhashis Ghoshal Jul 21 2025 09.00 - 12.30 Bayesian classification: methodology, algorithms and applications Subhashis Ghoshal will visit in July 2025 and present his work on Bayesian semi-supervised learning. The event will also feature short talks from the Schools of Maths and Informatics. Bayes Centre G.03
Jul 21 2025 09.00 - 12.30 Bayesian classification: methodology, algorithms and applications Subhashis Ghoshal will visit in July 2025 and present his work on Bayesian semi-supervised learning. The event will also feature short talks from the Schools of Maths and Informatics.
