This year, we met again in person with the Optimisation and AutoML community at COSEAL in Paris at Sorbonne University. Our former group member, Koen van der Blom, co-hosted the event. We were very happy to see him again!
COSEAL was well-attended, with over 100 registered participants from the AutoML groups in Freiburg and Hannover, the organising LIP6, the Slovenian Jozef Stefan institute, and even from local companies like Huawei France, among others. In other words, Little Ada met a lot of new friends. Of course, ADA was well-represented as well: Holger, Mitra, Samira, Julia and Hadar were there. We were happy to find many of our friends and collaborators at COSEAL, including Frank Hutter, Marius Lindauer, Lars Kotthoff, Saso Dzeroski and former ADA member Jeroen.
The first day, Julia gave her first academic talk on her work on AutoML for super-resolution, which she did for a master’s thesis. She also presented a poster on her current work on AutoML for methane plume detection (read more about it here). Hadar presented a poster on his master’s thesis work on Graph Neural Networks for SAT problems. The day ended on a (literal) high note, with a reception at the top of the tower of Sorbonne’s science faculty, with a beautiful view over Paris and, most notably, the Eifel tower.
The next day we had the opportunity to enjoy a long talk by Holger on the recent work on Neural Network Verification by Matthias and Annelot, which was rewarded with the best paper award at the AAAI Workshop on Safe AI. Samira presented yet another of our applications to AutoML to different fields by showcasing her work on AutoML for Radio Astronomy.
Finally, on Wednesday, Mitra gave a talk on AutoML for hybrid (physical) modelling in Earth Observation, a topic that Laurens and Nuno work on (Read more about Laurens’ work here, and Nuno’s work here [link upcoming]. Read the work by former master student Victor here).
After two and a half days of talks and presentations, it was time to say goodbye to COSEAL. Fortunately, some of us could stay to attend the AutoML workshop.
The AutoML workshop was a whole different event, focussed on discussions and exchanging ideas within the community. For two days, we had many fruitful discussions on topics including benchmarking, multi-objective optimisation, explainable and interactive AutoML, and ethics and fairness.
It was an exciting week, full of exchanges of research ideas and interesting discussions. A great success for ADA: we gave a peek into how varied our work in the field of AutoML is before returning to Leiden, a little tired but full of new ideas.