Seminars for the 2024 Season
This is will we will add our seminars as they take place. We conduct these all live, via Zoom call, then upload the recorded sessions as they occur. Please see schedule page for the dates for upcoming seminars.
Competition Kickoff
In our introductory session we kicked off the Fuzzy Explainable Fuzzy Challenge 2024. Topics included the overall competition layout, changes to the competition, and general information on how to register and communicate with the organizers.
Introduction to Asteroid Smashers
In our first seminar Dr. Tim Arnett introduced us to the game we will be using for this year's competition. Topics included how to get started with the competition version of Asteroid Smashers and how the compeition will be scored. We also dove in to a template we provide for your fuzzy controllers.
Introduction to Fuzzy Logic
In our second seminar Dr. Tim Arnett introduced us to the principles of Fuzzy logic and Fuzzy Inference Systems (FIS).
Gradient Free Optimization
In this seminar, we go over a number of different gradient-free optimization methods. Gradient-free methods are powerful in that they can be applied to a wide range of problems including both supervised and reinforcement learning problems. The talk is somewhat tailored towards Genetic Algorithms and their application to Fuzzy Systems - known as Genetic Fuzzy Systems.
Kessler Game Introduction
In this seminar, an intro to the Python game "Kessler" is given. Kessler is inspired by the classic arcade game "Asteroids" and will be the simulation environment used for this year's competition. Kessler is open source written in pure Python with machine/reinforcement learning in mind.
Explainable AI and Fuzzy Systems
In this talk, Dr. Scott Dick of the University of Alberta details XAI and its relationship with Fuzzy, its importance to the field of AI, and some of the criteria for explanations and their value as an interface between AI decision-making and humans.
Gradient Based Optimization for Fuzzy Systems
In this talk, Dr. Barnabas Bede details how to do gradient-based optimization of particular kinds of Fuzzy systems. These methods have applications in not only supervised learning, but also some reinforcement learning methods.
Reinforcement Learning with Fuzzy Systems
This talk, given by Dr. Barnabas Bede of Digipen Institute, details reinforcement learning both at a high level and its usage with Fuzzy systems. This ties in with his previous talk about gradient descent and can be valuable for developing your agent in Kessler.