PhD Research Students
MSc Data Science, University of Edinburgh, 2017; BSc Computer Science, Universitas Indonesia, 2015
Project: Opponent Modelling for Deep Multi-Agent Reinforcement Learning
The project aims to formulate efficient learning algorithms which enable agents to perform sequential decision making in multi-agent problems. The combination of opponent modelling techniques, which can provide additional information on opponent behaviour, and reinforcement learning algorithms are explored to achieve this aim. The project also investigates possible applications of the algorithms to multi-agent problems with high-dimensional observation space and partially-observable states.
MS in Computer Science, UC Davis, 2018; BS in Computer Science, UC Davis, 2016
Project: Re-envisioning Authentication by Behavioral Modeling of Autonomous Agents
I am designing a remote authentication scheme for computer networks using behavioral modeling of autonomous agents. Researchers in multi-agent theory have developed methods for predicting and modeling other such agents. These methods will be used to develop a novel authentication scheme with minimal security and privacy concerns, addressing deficiencies in modern protocols.
Diploma in Electronic and Computer Engineering, Technical University of Crete, 2017
Project: Safe Agent Modelling in Multi-Agent Learning
Multi-agent learning refers to how agents can learn and behave in the highly non-stationary multi-agent environment created by several agents that learn concurrently. Modeling the behavior of other agents is an important aspect of this research and focuses on predicting how other agents behave, react and adapt to the environment. Applications ranging from robotics to disaster response and the smart grid make use of agent modeling, and further advancements to this field will benefit them greatly. In our research, we focus on developing safe and efficient methods to explore agent behaviors.
Diploma in Electrical and Computer Engineering, Aristotle University of Thessaloniki, 2017
Project: Multi-Agent Reinforcement Learning and Communication in Partially Observable Environments
Multi-agent systems in partially observable environments face many challenging problems that traditional reinforcement learning algorithms fail to address. Agents have to deal with the lack of information about the environment's state and must cooperate in order to achieve the optimal reward. This project will investigate how existing reinforcement learning methods could be modified in order to train stable agents that are able to cooperate to solve complex tasks. Furthermore, using recent developments in deep learning, we will explore methods that enable agents to communicate in order to efficiently deal with the partial observability of environments.