Toward Human-guided Safe Reinforcement Learning in the Real World. This project aims to investigate human-guided safe reinforcement learning (RL). Safe RL is an important topic that could enable real applications of RL systems by addressing safety constraints. Existing safe RL assumes the availability of specified safety constraints in mathematical or logical forms. This project proposes to study learning safety objectives from information provided directly by humans or indirectly via language m ....Toward Human-guided Safe Reinforcement Learning in the Real World. This project aims to investigate human-guided safe reinforcement learning (RL). Safe RL is an important topic that could enable real applications of RL systems by addressing safety constraints. Existing safe RL assumes the availability of specified safety constraints in mathematical or logical forms. This project proposes to study learning safety objectives from information provided directly by humans or indirectly via language models, and human-guided continuous correction for safety improvements. The established theories and developed algorithms will advance frontier technologies in AI and contribute to a wide range of real applications of safe RL, such as robotics and autonomous driving, bringing enormous social and economic benefits. Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE240100144
Funder
Australian Research Council
Funding Amount
$444,447.00
Summary
Universal Model Selection Criteria for Scientific Machine Learning. This project aims to develop provably reliable universal model selection criteria to facilitate trustworthy scientific machine learning. Combining stochastic methods with an innovative geometric approach to basic statistical principles, this project expects to characterise, combine, and refine the most successful heuristics for designing and training huge models, such as deep neural networks, into a cohesive theoretical framewor ....Universal Model Selection Criteria for Scientific Machine Learning. This project aims to develop provably reliable universal model selection criteria to facilitate trustworthy scientific machine learning. Combining stochastic methods with an innovative geometric approach to basic statistical principles, this project expects to characterise, combine, and refine the most successful heuristics for designing and training huge models, such as deep neural networks, into a cohesive theoretical framework. The expected outcomes include a general toolkit for assisting neural network design at the forefront of scientific applications. This should significantly improve the quality of scientific predictions by facilitating confident adoption of deep learning methods into the pantheon of trustworthy modeling techniques. Read moreRead less