I am an applied mathematician. My research interests are in mathematical optimization. It is an area of mathematics that directly deals with the problem of making the best possible choice from a set of feasible choices. It seeks to understand how we achieve the best possible choice and how we can use this knowledge to improve management and technical decisions in science, engineering and commerce. Thinking in terms of choices is common in our cognitive culture and searching for the best possible choice is a basic human desire.Thus models of optimization arise everyday as management and technical decisions in many areas of human activity. I develop mathematical principles and methods for identifying and locating solutions of a wide range of optimization models.
PhD in Optimization, University of Melbourne
PRIZES AND AWARDS
2019: Joint winner of the 2019 Journal of Global Optimization Best Paper Prize, https://www.maths.unsw.edu.au/news/2019-11/jeya-guoyin-best-paper-award
2017: Winner of the 2015 Optimization Letters Best Paper Prize, https://link.springer.com/content/pdf/10.1007/s11590-016-1099-0.pdf
MY RECENT GRANTS
MOST CITED ARTICLES
I am interested in a range of topics in Optimization. One strand of my research examines optimization models in the face of data uncertainty. We examine mathematical approaches to finding robust best solutions of uncertain optimization models that are immunized against data uncertainty. We develop mathematical principles and methods for identifying and locating such solutions of a wide range of optimization models.
One paper, grown out our recent work in this area, won the Optimization Letters journal Best Paper Award for 2015.
A current project examines multi-objective optimization models in the presence of conflicting objectives and data uncertainty. It is a joint work with Dr Guoyin Li (Associate Professor), who is a member of the Optimization group at UNSW.
A second area of my research focuses on analysis, methods and applications of duality theory and associated techniques of convex optimization. We have examined a variety of topics such as the convex optimization models arising from the machine learning problems of large-scale data classification.
Some of my recent research projects in this area have involved technologically significant problems such as the development of screening algorithms for HIV-associated neurological disorders, produced by a new optimization procedure. This was a result of successful research collaboration between the research group in Optimization within Mathematics and the HIV Epidemiology and Clinical Research Group at St Vincent's hospital. The research outcome on screening algorithms was published in the British Journal, HIV Medicine.
What did HIV clinicians read on Medscape in 2011?
When HIV specialists come to Medscape HIV/AIDS what are they most interested in reading? ... Starting with number 10, here is our most-read content for 2011 in HIV. 10. Lifelong Antiretroviral Therapy Unsustainable, Experts Say. 9. Can We Stop Monitoring CD4 counts Entirely in HIV? ... 2. Curing HIV? And the number-one most-read article by HIV clinicians on Medscape in 2011 was...
A Screening Algorithm for HIV-Associated Neurocognitive Disorders, Cysique, Murray, Dunbar, Jeyakumar, Brew, HIV Medicine, 11:642-649, 2011. https://onlinelibrary.wiley.com/doi/full/10.1111/j.1468-1293.2010.00834.x
---SUSAN B COX, AUGUST 12 2011 WWW.MEDSCAPE.COM
Much of my studies in this area is of fundamental in nature, but these mathematical studies have a flow-through effect for addressing major challenges of modern problems such as the development of clinically based decision support tools in medical science. A current project is examining two-stage convex optimization with applications to medical decision-making optimization models of radiation therapy planning. Another on-going collaborative project examines applications of robust optimization for characterizing Huntington disease onset.
A third strand of my research looks at semi-algebraic global optimization problems, where the set of feasible choices is defined by polynomial equations and inequalities. These problems have numerous locally best solutions that are not globally best. We develop mathematical principles which can identify and locate the globally best solutions and can also be readily validated by computer methods using commonly available algorithms and software. A current project is examining bi-level global optimization models involving polynomials.
Nowadays, optimization together with sophisticated computer models is a widely used technology to improve performance in many industrial problems and emerging scientific applications. It is an interesting and challenging area of mathematical research. My current research project contains exciting components that are suitable for honours projects and PhD programs.
One study examines emerging applications of algebraic geometry to optimization over polynomials. A good understanding of convex sets in algebraic geometry will lead to insights into solving complex optimization problems involving polynomials.
Another project investigates bi-level optimization models. These models arise when two independent decision makers, ordered within a hierarchical structure, have conflicting objectives. Optimization models of this kind appear in resource allocation and planning problems.
A new project starting this year is examining optimization approaches to multi-stage optimization problems in the face of evolving uncertainty. The dynamic decision-making problems under uncertainty are multi-stage robust optimization problems, where uncertainties evolve over time (in stages), rendering traditional optimization solutions sub-optimal. This issue is particularly prevalent in medical decision-making, where a patient's condition can change during the course of the treatment. The aim of the project is to develop optimization principles to identify true optimal solutions of these multi-stage problems and develop associated methods to find those solutions.
Recently completed honours thesis topics include:
(i) Feature Selection under Data Uncertainty via DC Programming (2022)
(ii) Supervised Machine Learning under Data Uncertainty (2020)
(iii) Two-stage Adjustable Robust Optimization (2019)
(iv) Bilevel optimization and Principal-Agent Problems (2017)
(v) Portfolio Optimization under Data Uncertainty (2010)
(vi) Optimization approaches to simultaneous classification and feature selection (2007)
(vii) Semidefinite programming approaches to support vector machines (2003)
(viii) Optimization methods in data mining (2000)
I am interested in talented prospective postgraduate students/postdocs/honours students with interests in these areas.