The COnfiguration and SElection of ALgorithms (COSEAL) network of researchers regularly meets to discuss recent advances in the field of algorithm configuration and selection as well as all its related fields like Automated Machine Learning or general optimization.
Tobias Vente and Lukas Wegmeth presented “The Challenges of Algorithm Selection and Hyperparameter Optimization for Recommender Systems“, a summary of past advances and current research problems for Automated Recommender Systems, especially generalizability. Lennart Purucker presented “A First Look at Meta-Learning Algorithm Selection for Post Hoc Ensembling in AutoML“, a first study on learning to select the best algorithm for building an ensemble after model selection in AutoML systems.
The COSEAL Workshop attracts researchers from a wide variety of topics surrounding optimization, algorithm selection, meta-learning, automated machine learning, and many more. It is a great venue to discuss emerging ideas and topics with a large group of experts.
This year’s COSEAL Workshop was special in multiple ways. It was the tenth anniversary of the workshop, the first time in person since four years ago, and the number of attendees was more than twice the expected amount. The COSEAL Workshop 2023 sat on a tight schedule with numerous expert talks and a high volume of posters. It was the perfect environment for new insights and ideas.
Hosted at the Jussieu Campus of Sorbonne University in Paris, near the Cathédrale Notre-Dame and right next to Jardin des Plantes, we got to enjoy awesome sightseeing experiences and relaxing walks.
One highlight of the COSEAL23 workshop was the welcome reception on the top floor of Tour Zamansky. The sight of the whole city of Paris from up top at dusk and nightfall left every participant in awe. Besides the awesome view and the fantastic food, all participants could review the first day, discuss current topics and make new contacts.
We got the chance to meet with many experts from different fields, discussed ours and their research, and received valuable commentary and perspectives on our work. Besides many new faces, we were happy to see familiar researchers and engineers from the close-knit AutoML community again.
We want to thank the organizers: Carola Doerr, Elena Raponi, Anja Jankovic, Koen van der Blom, and Maria Laura Santoni. They worked hard to create this opportunity, and the workshop was a great success.