Partially Observable MDPs, Monte Carlo Methods, and Sustainable Fisheries. Partially Observable Markov Decision Processes (POMDPs) provide a general mathematical framework for sequential decision making under uncertainty. However, solving POMDPs effectively under realistic assumptions remains a challenging problem. This project aims to develop new efficient Monte Carlo algorithms to significantly advance the application of POMDPs to real-world decision problems involving complex action spaces an ....Partially Observable MDPs, Monte Carlo Methods, and Sustainable Fisheries. Partially Observable Markov Decision Processes (POMDPs) provide a general mathematical framework for sequential decision making under uncertainty. However, solving POMDPs effectively under realistic assumptions remains a challenging problem. This project aims to develop new efficient Monte Carlo algorithms to significantly advance the application of POMDPs to real-world decision problems involving complex action spaces and system dynamics. Both theoretical and algorithmic approaches will be applied to sustainable fishery management --- an important problem for Australia and an ideal context for POMDPs. The project will advance research in artificial intelligence, dynamical systems, and fishery operations, and benefit the national economy.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
Time consistency, risk-mitigation and partially observable systems. This project aims to find optimal decision rules that mitigate risk in a time consistent manner for partially observable systems. Many problems in conservation management and engineering systems are dependent on random environments and entail risk of failure. The challenge of consistently minimising such a risk while achieving satisfactory and sustainable resource consumption is considerable. This project aims to develop analyti ....Time consistency, risk-mitigation and partially observable systems. This project aims to find optimal decision rules that mitigate risk in a time consistent manner for partially observable systems. Many problems in conservation management and engineering systems are dependent on random environments and entail risk of failure. The challenge of consistently minimising such a risk while achieving satisfactory and sustainable resource consumption is considerable. This project aims to develop analytical and numerical methods for optimal control in such scenarios. These methods will have application to fishery management, communication networks, power systems and social resource allocation scenarios.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE130100291
Funder
Australian Research Council
Funding Amount
$374,595.00
Summary
Adaptive control of stochastic queueing networks. Queues of items competing for service appear on the road, in health-care, in manufacturing and in communication systems. This project will set up methodology for adaptive control and resource allocation for stochastic queueing network models applicable to a variety of scenarios accounting for parameter uncertainty.