28 July 2017
Dr Francois Petitjean, (Main supervisor)
Francois is a lecturer in the Monash Centre for Data Science. Francois is a highly awarded researcher these include an Australian Research Council, DECRA, a Victorian Young achiever award, an IBM Faculty award, the Deans award for Excellence in research impact and also for Excellence in research by an early career researcher.
- achieve excellence at a world top 100 university
- be part of an internationally recognised faculty of information technology
- become a world leader in analysing and making sense of the Sentinel-2 mission data
- contribute to the theory of machine learning
With the second satellite of the Sentinel-2 mission just launched in 2017, there is an incredible opportunity for the right student to become the world leader on how to analyse and make sense of this vast amount of data. It is anticipated that the project will make contributions to the theory of machine learning, with applications to the study of vegetation in general and more specifically in agriculture. The project however remains open if the successful candidate has other applications at heart (eg landslide, fire prediction). This project is fully funded. There is a paradigm shift in the way we can observe our planet: new-generation satellites (Sentinel-2, Landsat-8) are now imaging Earth completely, every 5 days, at high resolution, and at _no charge to end-users. It is not yet possible to tap the full value of this data, as existing machine learning methods for classifying time series cannot scale to such vast volumes of data. Temporal land-cover maps assign unique labels to geographic areas, describing their evolution over time. One of today’s key challenges is how to automatically produce these maps from the growing torrent of satellite data, to monitor Earth’s highly dynamic systems [a-h]. Presently, state-of-the-art research into time series classification lags behind the demands of the latest space missions, which produce terabytes of data each day. Why? Most of the research into time series classification has been done with datasets that hold no more than 10 thousand time series [i]. In contrast, the Sentinel-2 satellite gathers over 10 *trillion* time series, capturing Earth’s land surfaces and coastal waters at resolutions of 10-60m. Although much research has gone into classifying remote sensing images, few studies have analysed time series extracted from sequences of satellite images. This Project aims to create the machine learning technologies necessary to analyse series of satellite images, and to produce accurate temporal land-cover maps of our planet. Potential high-value applications for Australia include fire prevention, agricultural planning, and mining site monitoring and rehabilitation.
This PhD project is fully funded and includes a full tuition fee offset, as well as a stipend at RTP rate ($26,682 pa). Top-ups might be considered for outstanding students. You will be based in the Faculty of information Technology Monash University. Read more about Scholarships and Funding
This is incredible opportunity for the right student to become the world leader on how to analyse and make sense of this vast amount of data. It is anticipated that the project will make contributions to the theory of machine learning, with applications to the study of vegetation in general and more specifically in agriculture. The project however remains open if the successful candidate has other applications at heart (eg landslide, fire prediction).
Candidates need to be eligible to undertake a PhD in the Faculty of IT at Monash University. Please check your eligibility on the How to apply page and if you meet the criteria please submit an Expression of Interest.
Dr François Petitjean will be the main supervisor of this project. After receiving two prestigious awards for his PhD at the French Space Agency, François Petitjean decided that Big Data would be foundational to future scientific progress. He joined the Monash Centre for Data Science in 2012 and began developing new algorithms, proving how standard statistical methods can be applied to enormous datasets without sacrificing accuracy. Among numerous advantages, this work has unlocked the potential to develop more powerful and heat-resistant anti-inflammatory drugs, to discover new symptoms of rare diseases, to monitor oil spills, and to fight insect-borne diseases. The US Air Force and the Australian research Council are now funding further refinement of his methods to understand the evolution of our Planet.
Francois’s most recent articles:
- SDM 2017: Scalable time series classification - http://bit.ly/ScalableTSClassification
- Machine Leaning 2017: Discriminative parameter learning for BNs - http://bit.ly/DiscriminativeBNs
- KDD 2016: Multiple hypothesis testing - http://bit.ly/HypoTestKDD16
- DMKD 2016: Top-k sequential patterns - http://bit.ly/SkopusPaper
Additional relevant information:
Professor Geoff Webb, is a world-leading expert in Machine Learning and former Editor-in-Chief of the journal Data Mining and Knowledge Discovery. Geoff has published more than 200 scientific papers in the fields of machine learning, data science, data mining, data analytics, big data and user modeling. He is an editor of the Encyclopedia of Machine Learning. He created the Averaged One-Dependence Estimators machine learning algorithm and its generalization Averaged N-Dependence Estimators and has worked extensively on statistically sound association rule learning. His early work included advocating the use of machine learning to create black box user models; interactive machine learning; and one of the first approaches to association rule learning using minimum support and confidence. His awards include IEEE Fellow, Australian Computer Society ICT Researcher of the Year Award 2016, the IEEE International Conference on Data Mining Outstanding Service Award, an Australian Research Council Outstanding Researcher Award and multiple Australian Research Council Discovery Grants. See http://i.giwebb.com
Dr Christoph Rüdiger, is an expert in tracking physical variables over time from remote sensing data and will provide thematic expertise to the Project. Christoph has published 80+ scientific papers and received $3M+ of funding from Space agencies around the world and from the Australian Research Council. See http://users.monash.edu.au/~crudiger/
This project will be supported by the strong collaboration of the main supervisor with:
- The French Space Agency, for whom the main supervisor used to work for and has active collaboration with. The project includes funding for the PhD student to travel for 3 weeks to the French Space Agency.
- Agriculture Victoria and the VLUIS team, who is in charge of providing an land-cover map of Victoria (http://vro.agriculture.vic.gov.au/dpi/vro/vrosite.nsf/pages/vluis). If interested, possibilities exist for the successful student to participate to the data collection campaign around Victoria.
- Airbus and Thales; the main supervisor is a scientific advisor to the IRT St Exupery, which is a technology transfer institute founded by Thales, Airbus and the French Research Agency. There are possibilities of short-term internships within their teams depending on candidates' aspirations.
[a] J Inglada et al., “Assessment of an Operational System for Crop Type Map Production Using High Temporal and Spatial Resolution
Satellite Optical Imagery.” Remote Sensing, 2015.
[b] Sheffield, K., Morse-McNabb, E., Clark, R., Robson, S. and Lewis, H., 2015. Mapping dominant annual land cover from 2009 to 2013
across Victoria, Australia using satellite imagery. Scientific data, 2.
[c] F. Petitjean, J. Inglada and P. Gançarski, “Satellite Image Time Series Analysis under Time Warping,” IEEE Transactions on Geoscience
and Remote Sensing, vol. 50, no. 8, pp. 3081-3095, 2012.
[d] F. Petitjean, C. Kurtz, N. Passat & P. Gançarski, Spatio-Temporal Reasoning for the Classification of Satellite Image Time Series Pattern
Recognition Letters, 2012, Vol. 33, Num. 13, pp. 1805-1815.
[e] F. Petitjean, J. Inglada & P. Gançarski, “Assessing the quality of temporal high-resolution classifications with low-resolution satellite
images time series,” International Journal of Remote Sensing, 2014, Vol. 35, Num. 7, pp. 2693-2712.
[f] V Maus, et al. "Open boundary dynamic time warping for satellite image time series classification." IEEE IGARSS, 2015.
[g] X Guan et al. "Mapping Rice Cropping Systems in Vietnam Using an NDVI-Based Time-Series Similarity Measurement Based on DTW
Distance." Remote Sensing, 2016.
[h] R. Flamary et al. “Analysis of Multitemporal Classification Techniques for Forecasting Image Time Series,” IEEE Geoscience and
Remote Sensing Letters, 12.5 (2015): 953-957.
[i] Y. Chen, E. Keogh, B. Hu, N. Begum, A. Bagnall, A. Mueen and G. Batista (2015). The UCR Time Series Classification Archive. URL
Contact name: Dr Francois Petitjean