by Catherine Laflamme, RL Community Lead
An idea was born
The idea for an open community focused on reinforcement learning in Austria came from Catherine and Andreas last year, after realising that there were many of us working on similar topics – asking the same questions, confronted with the same problems, and looking for collaboration – but that we, for the most part, were working alone, without knowing what the others were doing.
Thus, the idea of the RL Community was born. The goal? Creating a community which enables the growth and progress of the field of reinforcement learning in Austria, from fundamental research to industry applications. The RL Community would provide a platform for networking, the exchange of ideas and expertise, establish partnerships between research institutions and scientific researchers and establish an ecosystem to support the progression of the field.
The RL Community would provide a platform for networking, the exchange of ideas and expertise, establish partnerships between research institutions and scientific researchers and establish an ecosystem to support the progression of the field.
Even AI isn’t immune to COVID-19
The first official event of the newly coined “RL Community” – established under the larger umbrella of AI Austria – was to be a meeting in Vienna, where interested members of the community could introduce themselves, their background, their experience in RL and get to know the others. It would also gain participant feedback to establish the directions of focus groups. Once the COVID-19 situation became clear, we knew we weren’t willing to wait for the meeting until it could happen in person, and thus the meeting format was adapted to be held online. So, in this way, on April 28,2020, 20+ reinforcement learning aspirants, enthusiasts and experts met via Web-Ex for the inaugural RL Community event!
Keynote from JKU Linz
After an introduction about AI Austria from Clemens, and an introduction to what the RL Community is all about from Andreas, the stage was set for the Keynote presentation from the JKU Linz. In this keynote we learned about the novel reinforcement learning approach “RUDDER”, an approach for delayed rewards in reinforcement learning scenarios, and shows improved results when tested on Atari games when compared to the baseline for delayed reward scenarios.
Participant Introduction
Even though we weren’t in person, the main goal of this first meeting was establishing a network among the participants. Therefore, each participant was given a 5 minute time slot to introduce themselves, their background and their experience with reinforcement learning. With participants from across Austria, ranging from university post-secondary students to industry experts, it provided an overview of just how much activity there is across the country.
Future Plans
The potential of reinforcement learning across application areas has just started to be recognized, and the RL community will work to provide an ecosystem to support and nurture the progress of the field here in Austria. This will be best achieved by combining forces of fundamental research and industry representatives, to find and develop solutions tailored to their needs.
For those interested in learning more and getting involved contact Catherine or Andreas under RL@aiaustria.com