Vamplew, Peter (Prof) - Research
Research interests and current research projects
My primary research focus over the last 15 years has been the extension of reinforcement learning methods to problems involving multiple conflicting objectives. While reinforcement learning has proven to be a very powerful means of creating autonomous AI systems, we contend that scalar rewards are inadequate for many real-world problems, and particularly for any attempts to create Artificial General Intelligence. Therefore my research has focused on the development, evaluation and application of more efficient multi-objective reinforcement learning algorithms (MORL).
More recently this work has extended to cover the application of MORL to addressing issues in creating human-aligned AI, including to areas such as explainability and safety of autonomous AI systems. This interest is reflected in my co-leadership of the Australian Responsible Autonomous Agents Collective (ARAAC) which spans several Australian universities, as well as my membership of the Future of Life Institute’s Existential AI Safety Research Community
Publications
For up-to-date publication details please refer to my online research profiles at Google Scholar and ORCID.
Research funding
I have been a member of investigatory teams receiving in excess of $4.2m in research funding support, including more than $650,000 as a lead or key investigator. Significant funded projects include:
- Analysis of Human-Machine Teams in Adversarial Environments, Air Force Office of Scientific Research and Defence Science and Technology Group Australian Autonomy Initiative
- Modelling Adversary Intent Using Multiobjective Reinforcement Learning, Science and Technology Group Operations Research Collaboration
- Investigating Multi-Agent Learning with Multiple and Uncertain Objectives, Defence Science and Technology Group AI for Decision-Making Initiative
- Multi-Objective Reinforcement Learning in Dynamic Team Based Adversarial Games, Science and Technology Group Scholarship (Project-Based) Funding Agreement
- Applying Machine Learning to identify Online Sexual Predators, Collaborative Research Program
- Multiobjective reinforcement learning for control of local battery storage for solar power systems, CSIRO Flagship Collaboration Fund
Graduate research supervisions
I have supervised more than twenty students to successful PhD and Masters completions in topics including artificial intelligence, reinforcement learning, video games and cybersecurity.
Current students:
Nik Goodger | PhD | Language representations for generalization in Reinforcement Learning. |
Haddie Harland | PhD | AI Apology: Beyond Explainable Reinforcement Learning |
Adrian Ly | PhD | Addressing the deadly triad in deep reinforcement learning |
Gavan Muller | PhD | Enabling Safe Behavior in Dynamic Environments through Responsibility Awareness |
Charlotte Young | PhD | Measuring the effectiveness of explanations of the decisions of artificial intelligence (AI) algorithms |
Past students:
Budi Kurniawan | 2022 | PhD | Single- and Multiobjective Reinforcement Learning in Dynamic Adversarial Games |
Alastair Lansley | 2020 | PhD | Adventures in software engineering : plugging HCI & acessibility gaps with open source solutions |
Ansam Khraisat | 2020 | PhD | Intelligent Zero-Day Intrusion Detection Framework for Internet of Things |
Paul Black | 2020 | PhD | Techniques for the reverse engineering of banking malware |
Ikram ul Haq | 2020 | PhD | Fraud detection for online banking for scalable and distributed data |
Adam Bignold | 2019 | PhD | Rule-based interactive assisted reinforcement learning |
Julie Ross | 2017 | PhD | Video game classification in Australia : Does it enable parents to make informed game choices for their children |
Nathan Robinson | 2016 | PhD | Assessing productive soil-landscapes in Victoria using digital soil mapping |
Rosemary Torney | 2015 | PhD | Application of psycholinguistic features to authorship profiling for first language, gender and age group |
Leigh Achterbosch | 2015 | PhD | Causes, magnitude and implications of Griefing in Massively Multiplayer Online Role-Playing Games |
Omaru Maruatona | 2014 | PhD | Internet banking fraud detection using prudent analysis |
Oana Ureche | 2014 | PhD | Static code analysis of data-driven applications through common lingua and the Semantic Web technologies |
Cameron Foale | 2011 | PhD | The Directional Propagation Cache: Real-time Acoustics Simulation for Immersive Computer Games |
Armita Zarnegar | 2011 | PhD | Regulatory network discovery using heuristics |
Mofakharul Islam | 2008 | Masters | Unsupervised Colour Texture Segmentation Using Markov Random Fields |
Subhasis Mukherjee | 2008 | Masters | Reinforcement Learning on the AIBO Robot |
Adam Berry | 2008 | PhD | Escaping the Bounds of Generality: Unbounded Biobjective Optimisation |
Robert Ollington | 2007 | PhD | A Biologically-Inspired Model for Robot Navigation |
Jonathan Pugh | 2001 | Masters | Multi-user virtual reality |
Uwe Rosebrock | 2000 | Masters | The influence of specific entry devices for navigation in virtual environments on the ability to gather spatial knowledge |
Carl Lewis | 1999 | PhD | The construction of training signals from incomplete information for use with sequential input classifiers |
Tim Freeman | 1998 | Masters | Application of Neural Network Classifiers to Electrocardiographic Body Surface Mapping |