IEEE Lone Star Section Joint SMC AESS Chapter meeting on February 20, 2020 to celebrate Engineers Week
Garrett Hall will present “Multi-Agent Reinforcement Learning in StarCraft II Environment.”
Reinforcement learning is a branch of machine learning which uses experiences to learn how to perform actions in an environment. The environment presented in this work is Star Craft II (SC2). SC2 is a real-time strategy game where two opponents use a variety of combat units and tactics to emerge victorious. Multi-agent reinforcement learning algorithms learn unique behavioral policies for each of their agents.
This presentation’s main focus will be the “Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning” algorithm known as QMIX. Using QMIX’s combination of recurrent neural networks and hypernetworks, agents act in a cooperative manner during skirmishes in the SC2 environment. After a demonstration of QMIX is given at the various stages of training, the presenter will discuss his current joint Exploratory IR and Thesis progress and its relationship to QMIX.
Please use the IEEE Lone Star Section registration site if you plan to attend. The meeting will be held in the Slick Café Private Dining Room #2 at Southwest Research Institute in San Antonio, Texas.
About the speaker:
Garrett Hall was born and raised in Austin, Texas. He served four years in the U.S. Air Force (USAF) as a
Medical Laboratory Technician. Upon receiving an honorable discharge from the USAF, he returned to school
and obtained an Associate Degree in Chemistry from Austin Community College. He went on to study Electrical
Engineering at the University of Texas at San Antonio (UTSA) where he graduated with a BSEE, Cum Laude.
He performed undergraduate research in the area of brain computer interfacing using machine learning. He is
currently working as an electrical engineer in the Avionics and Support Systems Department at Southwest Research
Institute and is pursuing an MSEE at UTSA with an emphasis on deep learning using adversarial examples and
multi-agent reinforcement learning.