STAI CDT Student, Nathan Gavenski, presented the demonstration track paper, ‘Imitation Learning Datasets: A Toolkit For Creating Datasets, Training Agents and Benchmarking’, at the 23rd International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2024). The conference took place in Auckland, New Zealand.
The paper discussed a toolkit that Nathan, and co-authors Professor Michael Luck and Dr Odinaldo Rodrigues, developed. The toolkit helps researchers create data more easily and efficiently, train imitation learning agents, and makes the process of comparing other imitation learning agents more straightforward. The toolkit is open source, which means that everyone can use and alter it to solve their research needs, and it is maintained by releasing new methods for other researchers to use and new datasets so researchers have access to the same data.
It was Nathan’s first in-person conference, as his Master’s was during COVID-19 and so it was an exciting opportunity for him to meet researchers working in the same area. He said, “I’m the only person doing research with imitation learning in STAI CDT and being able to meet other people working in the same field was great. There were few of us, but we had fun talking about how few of us were there!”.
AAMAS 2024 was also an opportunity to meet well-known researchers whose work Nathan had read. Most of the highlights of the conference for Nathan were the conversations he had with fellow researchers and also presenting the paper during the conference.
The conference also helped Nathan to feel more confident in his research. He said, “Receiving feedback from people excited to use the toolkit we developed in their research was quite a boost in how I feel about what I am doing here”.
Auckland was a beautiful setting for the conference and Nathan made the most of exporing the city, particularly by foot. He took part in group morning runs with fellow AAMAS 2024 attendees. As he explains, “One of the days, we ended up running to a very touristy place where you can see the entire city and the ocean. It was an amazing view and a very cathartic experience (we had to run up a hill!)”.
Congratulations to Nathan and we look forward to hearing about future conferences he attends in the future.
The full demonstration track paper can be found here: