FEIT Research Project Database

Reinforcement learning (stochastic optimisation and game theory)

Project Leader: Jonathan Manton
Primary Contact: Jonathan Manton (jmanton@unimelb.edu.au)
Keywords: machine learning
Disciplines: Electrical & Electronic Engineering

Whereas machine learning algorithms are given a set of test data to learn from, humans can learn by interacting with the environment. This ability to explore the environment, and test hypotheses, is known as reinforcement learning. This project will study reinforcement learning from the perspective of stochastic optimisation and game theory. Additional insight will be gained from creating and studying simulated environments with one or more "adult" agents and a "child" agent who must learn survival skills from the adult agents.