2018-10-Postdoctoral position in Applied Bayesian Statistics

  • Help
  • Find
  • Facebook
  • Twitter
E2S: Energy and Environment Solutions

What is E2S UPPA?

What is E2S UPPA?

The consortium at the heart of the Energy Environment Solutions (E2S) project is a composed of the University of Pau and the Pays de l’Adour (UPPA) and two national research organisations, National Institute...

Read more

You are here:

Postdoctoral position in Environmental Sciences

Scientific background / Project

Almost every area of human endeavour has been impacted in the past few decades by new forms of data. An immediate example is ‘big data’. In addition to a greater volume of data from traditional sources, there are also many more sources of potentially useful data. For example, traditional observational data can be more easily acquired using new technology, and can be complemented by new digital data acquired by sensors such as satellites, wearables and automatic monitoring devices. Complex data can also include ‘systems data’ which require a fusion of data sources to comprehensively describe a complicated process. Bayesian computational statistics offers an appealing framework for modelling and analysis of all of these forms of complex data. For example, models that encourage sparsity and adaptive sampling can help to address the problem of big data; informative priors can be employed to support little data; and Bayesian networks are an effective way of describing complicated systems.

The aim of this postdoc position is to develop and apply modern Bayesian methods to address the above challenges in the context of ecology and environmental sciences. Despite the intense international interest in statistical modelling of complex data, and of improved modelling of ecological and environmental data, there are still substantial gaps in both the statistical theory and methods and their application in this field. These gaps motive the current research proposal: to create new knowledge in Bayesian statistics to address problems in ecology and environmental sciences that involve complex data. The outputs of this research will benefit not only the fields of Bayesian statistics and environmental sciences, but will also translate to other areas of statistics and applied sciences.

The post doc will take part in the development of statistical methods for analyzing long-term ecological data and in statistical analyzes within the BIGCEES team (Big model and Big data in Computational Ecology and Environmental Sciences) based in Anglet (France).


Tasks and proposed methodology

Three main statistical research problems will be pursued.

  1. Bayesian methods for modelling and analysis of highly structured big data.
    This research problem will address the problem of modelling and analysis of complex data in the form of highly structured big data. In particular, the research will focus on spatio-temporal regression and variable selection for multivariate high dimensional data, which is a common but very challenging problem of international interest, with many applications in ecology, medicine, genomics and other areas.

  2. Bayesian methods for modelling and analysis of systems data.
    This research problem will address the problem of modelling and analysis of complexdata in the form of diverse data sources. In particular, the research will focus on Bayesian networks as an appealing graphical hierarchical approach for integrating diverse sources of data to probabilistically describe complicated systems, with particular attention on the appropriate quantification of uncertainty.

  3. Efficient algorithms for modelling and analysis of complex data.
    The models described above require highly efficient algorithms to implement, particularly given the size and complexity of the data collated from potentially multiple sources. This research project will focus on the compilation of appropriate algorithms, with the aim of creating an open repository for these types of problems.

The above statistical research problems will be applied to the following open problems in ecology and the environment.

  1. Using complex data to monitor the health of the Great Barrier Reef in Australia.
    The Great Barrier Reef is one of the natural wonders of the world. However, because it is over 2000km long, it is difficult to monitor the physical and biological condition of the reef, particularly when it’s under stress from forces such as cyclones and crown of thorns. There is strong interest in using new data sources such as remote sensing from satellites and drones, expert information from citizen science, and underwater images to fill the gaps. The methods derived above for high dimensional spatiotemporal multivariate data, elicitation of expert opinion, and Bayesian networks for species resilience, will all be directly applicable to this endeavour.

  2. Using complex data to detect anomalies in river water quality.
    Anomaly detection is a critical component of effective monitoring in a wide variety of areas such as cybersecurity, health, and ecology and environmental sciences. The project will focus on water quality monitoring using continuous water quality sensors in parallel. The multivariate measurements collected by each sensor are perturbed by missing data and inaccuracies or drift in measurement depending on the quality of the sensor. Methods will be developed for anomaly detection and correction in these water sensors, and assessment of the predictions of key water quality sensors. Networks of sensors will also be explored, with a view to developing new adaptive design-based  methods for sampling from the sensor data. The derived Bayesian methods will build on the above research into Bayesian networks, spatio-temporal analysis and use of informative priors about the nature of an anomaly.

Working conditions

Laboratory: LMA: Laboratory of Mathematics and its Applications,
UMR CNRS 5142 (https://lma-umr5142.univ-pau.fr/fr/index.html), Anglet (64600) France.

Scientific team: BIGCEES team
(Big model and Big data in Computational Ecology and Environmental Sciences)

Localisation: Université de Pau et des Pays de l’Adour, campus of Anglet, Pyrénées-Atlantiques, France

Starting period: December 2018 or January 2019

Duration: 1 to 3 years

Gross salary range: 2960 €/month
The salary of the successful candidate will be based on the level chart for teaching and research personnel in the salary system of French universities. The salary will be of 2960 euros/month (gross salary), including allowance for 64 hours teaching per year.

Funding: This post doc position is funded by the project E2S-UPPA (Energy Environment Solutions) which has a core scientific domain that focuses on Environment and Energy to meet challenges related to the energy transition, geo-resources, aquatic habitats and the environmental effects of natural and  anthropogenic changes.


Supervision and contact

Supervisory team:
Kerrie Mengersen (k.mengersen@qu.edu.au) and Benoit Liquet (benoit.liquet@univ-pau.fr)
at the LMA : Laboratory of Mathematics and its Applications, UMR CNRS 5142 (https://lma-umr5142.univ-pau.fr/fr/index.html) on the campus of Anglet (64600) France.
The LMA is one of the lab of the University of Pau and the Pays de l’Adour (UPPA), I-site laureate with its project E2S-UPPA.

For additional information and proposal, please contact: Pr Kerrie Mengersen Email: k.mengersen@qut.edu.au or Pr Benoit Liquet, Tel: + 33 6 95 46 10 61 Email: benoit.liquet @ univ-pau.fr


Young Researcher skills required

  • The applicant should have a Phd in Bayesian Statistics or a related field, with strong experience in developing and applying new statistical methods, and very good statistical programming skills.
  • The applicant should have some background in ecology or environmental science.
  • The applicant should have a strong publication record in relevant journals and a demonstrated record of research impact.
  • The applicants should have excellent communication skills.
  • The applicant should have experience in working in a cross-disciplinary team.

Application procedure

Applications must be sent as a single pdf file and submitted by email to benoit.liquet @ univ-pau.fr

They must include:

  • a cover letter addressing the skills required above (max 2 pages),
  • CV (max 2 pages)
  • a publication list
  • contact details of at least two relevant professionals who can provide a reference letter based on request
  • a copy of PhD diploma,
  • as well as the report provided after the PhD defense (‘Rapport de soutenance de thèse’ or equivalent) and reports from the principal examinators of the PhD defense jury (‘Avis des rapporteurs’ or equivalent) (optional!)

Application deadline

Please submit your application to  benoit.liquet @ univ-pau.fr, before October 31st, mentioning [Postdoc] in the subject of your email.