Associate Professor Scott Anthony Sisson

Associate Professor Scott Anthony Sisson

Fields of research: Statistics, Statistical Theory, Applied Statistics
Campus: Kensington
Tag: Expanding Knowledge

ABOUT ME

 

Biography

My research is in computational statistics. In particular, I enjoy working to try and solve "intractable" (computationally or analytically) statistical problems. This covers many areas, some of which are Bayesian statistics, big data, computational techniques and simulation-based algorithms, statistical extreme value theory (with environmental applications) and I cover a diverse range of scientific applications.

Education

  • PhD in  Statistics (Bristol University, UK,...

ABOUT ME

 

Biography

My research is in computational statistics. In particular, I enjoy working to try and solve "intractable" (computationally or analytically) statistical problems. This covers many areas, some of which are Bayesian statistics, big data, computational techniques and simulation-based algorithms, statistical extreme value theory (with environmental applications) and I cover a diverse range of scientific applications.

Education

  • PhD in  Statistics (Bristol University, UK, 2002)
  • MSc in Environmental Statistics and Systems (Lancaster University, UK, 1997)
  • BSc in Mathematics and Statistics (Lancaster University, UK, 1996)

 

RESEARCH

 

Research Goals

  • Have as much statistics fun as possible while playing with computers
  • Do cool things
  • Only tackle challenging problems
  • Try to avoid incremental research

Research in Detail

Bayesian Methods: Bayesian statistics represents uncertainty about observed data, model parameters and even the model through probability distributions. As a consequence, this uncertainty is naturally incorporated into any statistical inference. Interest in Bayesian methods has exploded since the early 1990s thanks to the development of cheap computational power.

Big Data: As the size of datasets requiring analysis increases, so must the statistical techniques used to analyse them be able to efficiently handle the increase in scale. Standard statistical approaches, both classical and Bayesian, were not designed with this in mind. We are interested in developing ways to improve existing statistical techniques, as well as developing completely new methods, to better cope with big data.

Computational Statistics and Algorithms: Bayesian statistics depends on sophisticated simulation algorithms, such as Markov chain Monte Carlo, sequential Monte Carlo and many others. We are interested in developing new and more efficient algorithms in general and specific cases. We are also interested in developing new statistical approaches for models with computationally intractable likelihood functions, such as Bayesian "likelihood-free" methods (e.g. approximate Bayesian computation), indirect inference and composite likelihoods.

Extreme Value Theory: Extreme value theory involves the construction of methematical models that describe the extreme levels of real world processes. Our focus is the development of new spatial and temporal models for extremes (such as max-stable and max-infinitely-divisible models), and new statistical procedures for fitting them, as many models for extremes tend to be difficult to work with computationally. We are particularly interested in the extremes of climate and environmental processes.

Diverse Applications: We are interested in using Bayesian methods to solve challenging scientific problems in a broad range of disciplines. These include biology, climate science, ecology, hydrology and population genetics among others. We always work closely with experts in each field to obtain the best possible results. Our applied research occasionally generates press coverage.

Research Grants

  • Sustained funding from the Australian Research Council

Current Student Projects (PhD)

Short summary/title – relatively specific, but if you have lots of projects then write area summaries. This section allows students to judge the size and dynamic of the lab they might work in.

Boris Beranger - Statistical extreme value theory

Stephanie Clark - Self organising maps

Guilherme Rodrigues - Approximate Bayesian computation

Huaxin Xu - Particle algorithms on GPUs

Xin Zhang - Statistical methods in big data

Supervision Opportunities/Areas

Project supervision is available in all of the above research areas.

Advice for prospective students

Understand what you enjoy doing, and a broad idea of the kind of thing that you would like to do. Then find the right people to make it happen!

 

TEACHING & OUTREACH

 

Courses I teach

These vary each session.

Professional affiliations and service positions

Associate Editor: Australia and New Zealand Journal of Statistics

Associate Editor: Electronic Journal of Statistics

Associate Editor: Journal of the Royal Statistical Society, Series C

Associate Editor: STAT

Guest Editor: Computational Statistics and Data Analysis, 2015 Special Issue on Bayesian Computation

Bayesian Section Chair of the Statistical Society of Australia

President of the Australasian Society of Bayesian Analysis (a local chapter of ISBA)

Past President of the NSW Branch of the Statistical Society of Australia (2012-13)

AWARDS & ACHIEVEMENTS

 

2011  Moran Medal, joint with Dr. M. M. Tanaka (Australian Academy of Science)

2010  J. G. Russell Award (Australian Academy of Science)

2010 Queen Elizabeth II Research Fellowship (Australian Research Council)

2006  John Yu Fellowship to Europe (Imperial College, London)

2005  UNSW Faculty of Science Learning and Teaching Award 

Location

School of Mathematics and Statistics
University of New South Wales
Sydney NSW 2052
The Red Centre
Room 1034


 

Map reference

Contact

9385 7027
9385 7123