A Novel Approach to Semi-Supervised Statistical Machine Learning. Recent successes in the construction of classifiers for making diagnoses and predictions are due in part to their using much data labelled with respect to their class of origin. But typically there are little labelled data but plentiful unlabelled data. The goal of semi-supervised learning (SSL) is to leverage large amounts of unlabelled data to improve the performance using only small labelled datasets and so SSL is of paramount ....A Novel Approach to Semi-Supervised Statistical Machine Learning. Recent successes in the construction of classifiers for making diagnoses and predictions are due in part to their using much data labelled with respect to their class of origin. But typically there are little labelled data but plentiful unlabelled data. The goal of semi-supervised learning (SSL) is to leverage large amounts of unlabelled data to improve the performance using only small labelled datasets and so SSL is of paramount importance to applications where it is expensive or impractical to obtain much labelled data. The project is to develop a novel SSL approach that adopts a missingness mechanism for the missing labels to build a classifier that not only improves accuracy but it can be greater than if the missing labels were known.
Read moreRead less
Stochastic majorization--minimization algorithms for data science. The changing nature of acquisition and storage data has made the process of drawing inference infeasible with traditional statistical and machine learning methods. Modern data are often acquired in real time, in an incremental nature, and are often available in too large a volume to process on conventional machinery. The project proposes to study the family of stochastic majorisation-minimisation algorithms for computation of inf ....Stochastic majorization--minimization algorithms for data science. The changing nature of acquisition and storage data has made the process of drawing inference infeasible with traditional statistical and machine learning methods. Modern data are often acquired in real time, in an incremental nature, and are often available in too large a volume to process on conventional machinery. The project proposes to study the family of stochastic majorisation-minimisation algorithms for computation of inferential quantities in an incremental manner. The proposed stochastic algorithms encompass and extend upon a wide variety of current algorithmic frameworks for fitting statistical and machine learning models, and can be used to produce feasible and practical algorithms for complex models, both current and future.
Read moreRead less
Large Markov decision processes and combinatorial optimisation. Markov decision processes continue to gain in popularity for modelling a wide range of applications ranging from analysis of supply chains and queueing networks to cognitive science and control of autonomous vehicles. Nonetheless, they tend to become numerically intractable as the size of the model grows fast. Recent works use machine learning techniques to overcome this crucial issue, but with no convergence guarantee. This project ....Large Markov decision processes and combinatorial optimisation. Markov decision processes continue to gain in popularity for modelling a wide range of applications ranging from analysis of supply chains and queueing networks to cognitive science and control of autonomous vehicles. Nonetheless, they tend to become numerically intractable as the size of the model grows fast. Recent works use machine learning techniques to overcome this crucial issue, but with no convergence guarantee. This project aims to provide theoretically sound frameworks for solving large Markov decision processes, and exploit them to solve important combinatorial optimisation problems. This timely project can promote Australia's position in the development of such novel frameworks for many scientific and industrial applications.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE200100200
Funder
Australian Research Council
Funding Amount
$418,398.00
Summary
Next generation causal inference methods for biological data. This project aims to develop next generation causal inference methods for analysing biological data especially the single cell sequencing data and their applications in cell biology. Although Artificial Intelligence and Statistical Machine Learning have been applied successfully in many fields, including biological research, there is still a serious lack of methods for interpreting and reasoning about the mechanism of biological syste ....Next generation causal inference methods for biological data. This project aims to develop next generation causal inference methods for analysing biological data especially the single cell sequencing data and their applications in cell biology. Although Artificial Intelligence and Statistical Machine Learning have been applied successfully in many fields, including biological research, there is still a serious lack of methods for interpreting and reasoning about the mechanism of biological systems, the ultimate goal of research in many areas. Efficient data-driven causality discovery approaches developed by the project will be a timely and significant contribution to the knowledge of biology and statistics as well as the battle against health threats.
Read moreRead less
Perturbations in Complex Systems and Games. This project aims to: advance the perturbation theory of dynamic and stochastic games; further develop approximations of infinite dimensional linear programs by their finite dimensional counterparts, and by finding asymptotic limits of spaces of occupational measures, by solution of successive layers of fundamental equations; explain and quantify the "exceptionality" of instances of systems that are genuinely difficult to solve; and, capitalise on the ....Perturbations in Complex Systems and Games. This project aims to: advance the perturbation theory of dynamic and stochastic games; further develop approximations of infinite dimensional linear programs by their finite dimensional counterparts, and by finding asymptotic limits of spaces of occupational measures, by solution of successive layers of fundamental equations; explain and quantify the "exceptionality" of instances of systems that are genuinely difficult to solve; and, capitalise on the outstanding performance of our Snakes-and-Ladders Heuristic (SLH) for the solution of the Hamiltonian cycle problem to identify its "fixed complexity orbits" and generalise this notion to other NP-complete problems.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE210101352
Funder
Australian Research Council
Funding Amount
$330,000.00
Summary
Inverting the Signature Transform for Rough Paths and Random Processes. The signature transform provides an effective summary of the essential information encoded in multidimensional paths that are highly oscillatory and involve complicated randomness. The main goal of this project is to develop new algorithmic methods to reconstruct rough paths and random processes from the signature transform at various quantitative levels. This project expects to make theoretical breakthrough on the significa ....Inverting the Signature Transform for Rough Paths and Random Processes. The signature transform provides an effective summary of the essential information encoded in multidimensional paths that are highly oscillatory and involve complicated randomness. The main goal of this project is to develop new algorithmic methods to reconstruct rough paths and random processes from the signature transform at various quantitative levels. This project expects to make theoretical breakthrough on the significant open problem of signature inversion, thereby advancing knowledge in the areas of rough path theory and stochastic analysis. The newly developed methods will be utilised in combination with the emerging signature-based approach to study important problems in financial data analysis and visual speech recognition.
Read moreRead less