Here some of the talks that I gave during my master studies.
The perks of creating diverse agents in reinforcement learning
In this talk I give a short overview of methods to create a diverse set of reinforcement learning agents, especially a hypernetwork
approach.
Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model
Here I present the key concepts of the popular MuZero paper.
Deep Ensembles : A Loss Landscape Perspective
A talk about the advantages of creating diverse network ensembles to enhance generality.
Effective Diversity
A presentation about methods to effecivly generate a diverse set of agents.