ETH Zürich
Robert Meier
Institut für Theoretische Informatik
OAT Z 11
Andreasstrasse 5
8050 Zürich
E-Mail: romeier@inf.ethz.ch
I am a third-year Ph.D. student in Angelika Steger's group. My main research interest is Reinforcement Learning. Currently, I am working on unsupervised, open-ended, and (automated) curriculum learning algorithms. Due to my background as a mathematics student, I am also interested in understanding the algorithms we employ. The recent developments in computation speed allow a more thorough empirical evaluation of the learning behavior of agents. If you are a student interested in these topics and looking for a thesis topic, feel free to contact me.
Online learning of long-range dependencies
joint* with N. Zucchet*, S. Schug*, A. Mujika and João Sacramento (*equal contribution)
arxiv preprint (under review), 2023
A Simple Optimal Algorithm for the 2-Arm Bandit Problem
joint with M. Larcher and A. Steger
SIAM Symposium on Simplicity in Algorithms, 2023
Random initialisations performing above chance and how to find them
joint with F. Benzing, S. Schug, J. v. Oswald, Y. Akram, N. Zucchet, L. Aitchison, and A. Steger
14th International OPT Workshop on Optimization for Machine Learning, 2022. (Poster)
Open-Ended Reinforcement Learning with Neural
Reward Functions
joint* with A. Mujika* (* equal contribution))
36th Conference on Neural Information Processing Systems, 2022. (Poster)
Previously at the Agent Learning in Open-Endedness Workshop at ICLR 2022
An optimal
decentralized (Δ+1)-coloring algorithm
joint with D. Bertschinger, J. Lengler, A. Martinsson, A.
Steger, M. Trujić, and E. Welzl
28th Annual European Symposium on Algorithms, 2020
Simple Algorithms for the Multi-Armed Bandit Problem, Swiss Computational Neuroscience Retreat 2023