Discovery Early Career Researcher Award - Grant ID: DE170101134
Funder
Australian Research Council
Funding Amount
$360,000.00
Summary
Feasible algorithms for big inference. This project aims to develop algorithms for computationally-intensive statistical tools to analyse Big Data. Big Data is ubiquitous in science, engineering, industry and finance, but needs special machine learning to conduct correct inferential analysis. Computational bottlenecks make many tried-and-true tools of statistical inference inadequate. This project will develop tools including false discovery rate control, heteroscedastic and robust regression an ....Feasible algorithms for big inference. This project aims to develop algorithms for computationally-intensive statistical tools to analyse Big Data. Big Data is ubiquitous in science, engineering, industry and finance, but needs special machine learning to conduct correct inferential analysis. Computational bottlenecks make many tried-and-true tools of statistical inference inadequate. This project will develop tools including false discovery rate control, heteroscedastic and robust regression and mixture models, via Big Data-appropriate optimisation and composite-likelihood estimation. It will make open, well-documented, and accessible software available for the scalable and distributable analysis of Big Data. The expected outcome is a suite of scalable algorithms to analyse Big Data.Read moreRead less
Rating and ranking sports players and teams using minimum message length. All sorts of games and sports could use better systems for rating and ranking teams. This is as true in sports-mad Australia as any other country. Improved and more accessible rating systems across a variety of activities should encourage the general public to take a greater interest in the mathematics, statistics, information theory and machine learning behind the systems. With Cadability as our Australia-based interna ....Rating and ranking sports players and teams using minimum message length. All sorts of games and sports could use better systems for rating and ranking teams. This is as true in sports-mad Australia as any other country. Improved and more accessible rating systems across a variety of activities should encourage the general public to take a greater interest in the mathematics, statistics, information theory and machine learning behind the systems. With Cadability as our Australia-based international industry partner, the global use of these systems will be to Australia's economic advantage. Having a more accurate rating system which is wider-reaching both in the number of sports and games and the number of participants per sport and game should also encourage greater participation from the general public.Read moreRead less
Bayesian estimation of flexible spatial models with applications in medical imaging and econometric modeling. This project aims to develop statistical methodology for estimating flexible highly parameterised Bayesian spatial models. The flexible models examined will include regression, choice and time series models for data that is spatially registered. Spatial smoothing of parameters in the models will involve application of hierarchical spatial prior distributions. The resulting methodology wi ....Bayesian estimation of flexible spatial models with applications in medical imaging and econometric modeling. This project aims to develop statistical methodology for estimating flexible highly parameterised Bayesian spatial models. The flexible models examined will include regression, choice and time series models for data that is spatially registered. Spatial smoothing of parameters in the models will involve application of hierarchical spatial prior distributions. The resulting methodology will be applied to the analysis of medical imaging data and to the estimation of spatial econometric models of residential real estate prices. The expected outcomes include developments in the frontier framework of Bayesian computational estimation methodology, improved methods for medical image processing and estimation of high resolution spatial models of residential real estate prices in Australian metropolitan centres.Read moreRead less
Machine learning in adversarial environments. Machine learning underpins the technologies driving the economies of both Silicon Valley and Wall Street, from web search and ad placement, to stock predictions and efforts in fighting cybercrime. This project aims to answer the question: How can machines learn from data when contributors act maliciously for personal gain?
Australian Laureate Fellowships - Grant ID: FL150100150
Funder
Australian Research Council
Funding Amount
$2,413,112.00
Summary
Bayesian learning for decision making in the big data era. Bayesian learning for decision making in the big data era: This fellowship project aims to develop new techniques in evidence-based learning and decision-making in the big data era. Big data has arrived, and with it a huge global demand for statistical knowledge and skills to analyse these data for improved learning and decision-making. This project will seek to address this need by creating a step-change in knowledge in Bayesian statist ....Bayesian learning for decision making in the big data era. Bayesian learning for decision making in the big data era: This fellowship project aims to develop new techniques in evidence-based learning and decision-making in the big data era. Big data has arrived, and with it a huge global demand for statistical knowledge and skills to analyse these data for improved learning and decision-making. This project will seek to address this need by creating a step-change in knowledge in Bayesian statistics and translating this knowledge to real-world challenges in industry, environment and health. The new big data statistical analysts trained through the project could also create much needed capacity at national and international levels.Read moreRead less
Inverse and related problems in statistics. Modern statistical inverse problems arise in fields from astronomy and biology to engineering and finance. Sometimes the problems involve the analysis of small samples of very high dimensional data, and are central to information aquisition in areas such as genomics and signal analysis. All these topics are of significant national importance, and their solution will bring national and community benefits. In addition, the program to which the proposa ....Inverse and related problems in statistics. Modern statistical inverse problems arise in fields from astronomy and biology to engineering and finance. Sometimes the problems involve the analysis of small samples of very high dimensional data, and are central to information aquisition in areas such as genomics and signal analysis. All these topics are of significant national importance, and their solution will bring national and community benefits. In addition, the program to which the proposal will lead will be used extensively for research training. In Australia, where the demand for research-trained statisticians greatly exceeds supply, this contribution to the nation and the community will be particularly important. Read moreRead less
New and computationally feasible methods of constructing efficient and exact confidence limits from count data. Biological and health science data is commonly in the form of counts. The statistical analysis of such data should be (a) efficient i.e. it should not, in effect, throw away valuable data, (b) exact i.e. it should have precisely known statistical properties and (c) computationally feasible. Kabaila and Lloyd (1997-2001) have proposed and analysed a radically new method of confidence li ....New and computationally feasible methods of constructing efficient and exact confidence limits from count data. Biological and health science data is commonly in the form of counts. The statistical analysis of such data should be (a) efficient i.e. it should not, in effect, throw away valuable data, (b) exact i.e. it should have precisely known statistical properties and (c) computationally feasible. Kabaila and Lloyd (1997-2001) have proposed and analysed a radically new method of confidence limit construction which, for the first time, possesses all of these requirements. The purpose of the project is to establish further theoretical support for the new method, to develop efficient computational algorithms and to write easy-to-use computer programs for its practical use.Read moreRead less
Classification methods for providing personalised and class decisions. This project provides a novel approach to the clustering of multivariate samples on entities in a class that automatically matches the sample clusters across the entities, allowing for inter-sample variation between the samples in a class. The project aims to develop a widely applicable, mixture-model-based framework for the simultaneous clustering of multivariate samples with inter-sample variation in a class and for the mat ....Classification methods for providing personalised and class decisions. This project provides a novel approach to the clustering of multivariate samples on entities in a class that automatically matches the sample clusters across the entities, allowing for inter-sample variation between the samples in a class. The project aims to develop a widely applicable, mixture-model-based framework for the simultaneous clustering of multivariate samples with inter-sample variation in a class and for the matching of the clusters across the entities in the class. The project will use a statistical approach to automatically match the clusters, since the overall mixture model provides a template for the class. It will provide a basis for discriminating between different classes in addition to the identification of atypical data points within a sample and of anomalous samples within a class. Key applications include biological image analysis and the analysis of data in flow cytometry which is one of the fundamental research tools for the life scientist.Read moreRead less
Statistical methods for quantifying variation in spatiotemporal areal data. This project aims to develop new statistical methods for extracting insights into spatial and temporal variation in areal data. These tools will extend the Australian Cancer Atlas which provides small area estimates for 20 cancers across Australia. The project is significant because it will allow government and other organisations to reap dividends from investment in collecting spatial information and it will enable mode ....Statistical methods for quantifying variation in spatiotemporal areal data. This project aims to develop new statistical methods for extracting insights into spatial and temporal variation in areal data. These tools will extend the Australian Cancer Atlas which provides small area estimates for 20 cancers across Australia. The project is significant because it will allow government and other organisations to reap dividends from investment in collecting spatial information and it will enable modelled small-area estimates to be released without compromising confidentiality. The expected outcomes include new statistical knowledge and new insights into cancer. The results will benefit the many disciplines, managers and policy makers that make decisions based on geographic data mapped over space and time. Read moreRead less
Flexible Models and Methods for Longitudinal Data. The availability of increasingly large data sets offers the potential to improve understandings of many phenomena. However, without models for these phenomenon and methods to analyse the data generated by them, information contained in such data cannot be extracted. This project aims to advance statistical methods and models for analysing data that are collected on a large number of individuals at many time points. In particular, data collected ....Flexible Models and Methods for Longitudinal Data. The availability of increasingly large data sets offers the potential to improve understandings of many phenomena. However, without models for these phenomenon and methods to analyse the data generated by them, information contained in such data cannot be extracted. This project aims to advance statistical methods and models for analysing data that are collected on a large number of individuals at many time points. In particular, data collected from mobile phone applications will be used to understand the effect that training regimes have on cognitive functioning and how these effects vary with individual characteristics.Read moreRead less