果冻传媒

Skip to main content Skip to navigation

Statistical Learning & Inference Seminars

The seminars will take place every Tuesday 11am-12pm during term time.

Special Pre-term seminar

Date, Time and Room Speaker Title

24/09, 11am, B3.02 (Zeeman)

Vecchia approximation for Deep Gaussian Processes
Abstract: Deep Gaussian processes have several advantages compared to standard Gaussian Processes (GPs), including learning local and compositional structures of the signal. However, their practical applicability is hindered by their computational complexity and instability of approximation methods. In this work we propose a novel Vecchia approximation of Deep GPs, derive optimal contraction rates for structured (compositional) functions and discuss algorithmic aspects. We also demonstrate the applicability of the proposed method on synthetic data sets. Based on an ongoing joint work with Ismael Castillo (Sorbone, Paris), Thibault Randrianarisoa (Vector Institute, Toronto) and Yichen Zhou (University of Hong Kong).

Term 1, 26-27

Date, Time and Room Speaker Title
06/10, 11am, MB2.22  
Abstract:  
13/10, 11am, MB2.22  
Abstract:  
20/10, 11am, MB2.22   TBC  
Abstract:  
27/10, 11am, MB2.22

 
Abstract:  
03/11, 11am, MB2.22

 
Abstract:  
10/11, 11am, MB2.22 Bayesian Computation for Spatial Comparative Judgement Models

Abstract:

Comparative judgement studies elicit quality assessments through pairwise comparisons, typically analysed using the Bradley–Terry model. We have developed a suite of spatial comparative judgement models and run studies to map the prevalence of crimes such as domestic abuse, modern slavery, and forced marriage. This talk describes work on extracting more information per comparison, organised around three extensions to the standard model. First, a multivariate normal prior distribution whose covariance matrix is built from the network of areas in the study. This lets neighbouring areas borrow strength and reduces the data required by around ninety per cent. Second, a P贸lya-Gamma latent variable representation that yields a Gaussian full conditional while retaining that correlated prior distribution, which allows inference to be carried out in 20 to 30 seconds. Third, an experimental design that weights pairs by the prior variance they explain, made computable for large studies by a reduced basis decomposition of the associated pairs-of-pairs covariance matrix. We demonstrate these methods in several real world studies, describe the changes they have brought about in safeguarding practice, and outline our steps towards commercialising the work.
17/11, 11am, MB2.22 TBC  
Abstract:  
24/11, 11am, MB2.22  
Abstract:  
01/12, 11am, MB2.22

 
Abstract:  
08/12, 11am, MB2.22    
Abstract:  

Term 2, 26-27

Date, Time and Room

Speaker

Title

12/01, 11am, MB0.08

Anders Kock (Oxford)

 
Abstract:  
19/01, 11am, MB2.22

Kartik Waghmare (ETH Zurich)

 
Abstract:  
26/01, 11am, MB2.22 Alex Gibberd (Heriot-Watt)  
Abstract:  
02/02, 11am, MB2.22 Chao Zheng (Southampton)  
Abstract:  
09/02, 11am, MB2.22    
Abstract:  
16/02, 11am, MB2.22 Qingyuan Zhao (Cambridge)  
Abstract:  
23/02, 11am, MB2.22 Oliver Dukes (Ghent) TBC  
Abstract:  

02/03, 11am, MB2.22

   
Abstract:  
09/03, 11am, MB2.22    
Abstract:  
16/03, 11am, MB2.22    
Abstract:  

Term 3, 26-27

Date,Time and Room

Speaker Title

27/04, 11am MB0.07

   

Abstract:

 

04/05, 11am MB0.07

   

Abstract:

 

11/05, 11am MB0.07

   

Abstract:

 

18/05, 11am MB0.07

   

Abstract:

 

25/05, 11am MB0.07

   

Abstract:

 

01/06, 11am MB0.07

   

Abstract:

 

08/06, 11am MB0.07

   

Abstract:

 

15/06, 11am MB0.07

   

Abstract:

 

22/06, 11am MB0.07

   
Abstract:  

29/06, 11am MB0.07

   

Abstract:

 

Past seminars of this series

2025/26 seminars

2024/25 seminars

2023/24 seminars

Let us know you agree to cookies