Evolutionary Imitation Learning for Continous Control

Scope: Master’s Project/Master’s thesis

Advisor: Dr. Kevin S. Luck (kevin.s.luck@aalto.fi)

Added: 31.August 2022

Topic: Many recent works have sought to combine evolutionary strategies and algorithms with deep reinforcement learning to combine gradient-based and gradient-free optimization. This research thesis will investigate further potentials for the application of evolutionary strategies to the problem of imitation learning in a low-data regime, with multiple possible routes and the potential to identify and develop your own research problems. The thesis will start with an initial literature review to identify the space of potential hypothesis to investigate and apply the developed method to continous control tasks. This thesis is well suited for students interested in a research oriented master thesis with some possibilities to develop your own idea. For further information, contact the supervisor Kevin Luck.

Minimum knowledge:

  • Python
  • Some knowledge in Machine Learning

Preferred knowledge:

  • Has taken classes in Machine Leanring or Reinforcement Learning
  • Linux Skills
  • Has used frameworks such as pytorch

Thesis student will:

  • learn to develop and conduct research projects
  • learn the current state-of-the-art in reinforcement learning and evolutionary optimization
  • learn to handle learning frameworks such as pytorch

Related Literature: