Dr Joab Winkler
PhD
School of Computer Science
Reader
MSc Admissions Tutor
City College Liaison
Member of the Machine Learning research group
+44 114 222 1834
Full contact details
School of Computer Science
Regent Court (DCS)
211 Portobello
Sheffield
S1 4DP
- Profile
-
Joab Winkler is a Reader in The Department of Computer Science at The University of Sheffield. He obtained his undergraduate and PhD degrees at Imperial College London and University College London, respectively. He worked for a few years in industry, before returning to university to conduct research into algebraic and numerical properties of curves and surfaces in computer-aided design systems. He has extended this interest in numerical methods to work on blind image deconvolution, optical flow and neural networks. More details are in the section 'Research interests'.
- Research interests
-
Joab Winkler’s main research interest is the algebraic and numerical properties of curves and surfaces in computer-aided design systems. Most of this work has been performed using resultant matrices, and this has led him to consider more general issues of robust computations on polynomials that are corrupted by added noise. Examples include the computation of a structured low rank approximation of the Sylvester resultant matrix, the deconvolution of two polynomials and the determination of an approximate greatest common divisor of two polynomials.
He has developed a polynomial root solver for the determination of multiple roots of the theoretically exact form of a polynomial, when the coefficients of the given polynomial are corrupted by added noise. He has applied the methods used in this work on polynomials to the deblurring of an image when the point spread function is unknown and must therefore be computed. This is a difficult problem because it is necessary to determine the rank of a matrix whose entries are subject to error. This is one of the most difficult problems in linear algebra.
He has also considered numerical and computational issues in optical flow, which is an important problem in computer vision, in order to identify problems that arise in two commonly used algorithms.
More recently, he has started work on numerical properties of echo state networks, which are one form of a neural network.
- Publications
-
There has been a problem showing this information. Please try again later.
- Grants
-
Research Grants
- Sparse linear models: Their existence and stability, EPSRC, 09/2022 - 07/2023, £79,384, as PI
- Travel Grant, EPSRC, 07/2001 to 08/2001, £1,430, as PI
- Tensor Tomography for the Three Dimensional Photoelasticity, EPSRC, 11/2002 to 04/2006, £38,818, as PI
- Robust Computations in Geometric Modelling, EPSRC, 02/2005 to 01/2008, £71,785, as PI
- Travel Grant, EPSRC, 10/2005 to 01/2006, £4,100, as PI
- Polynomials and geometric modelling, ROYAL ACADEMY OF ENGINEERING (THE), 01/2010 to 12/2012, £23,613, as PI
- Professional activities and memberships
-
- Awarded an EPSRC Advanced Research Fellowship (1995)
- Co-organiser of a workshop, supported by the EPSRC, on The Representation and Management of Uncertainty in Geometric Computations (2001)
- Co-organiser of The Sheffield Machine Learning Workshop, supported by the EPSRC (2004)
- Organised the Summer School, supported by the EPSRC, Solving Polynomial Equations and Structured Matrix Methods for Approximate GCD Computations (2007)
- Awarded a Global Research Award by The Royal Academy of Engineering (2010-2011)
- Co-organiser of a conference in Kalamata, Greece on structured methods in numerical linear and multilinear algebra (2014)