Data Challenges 2023 (2)

Abstract

Challenges are proposed by public services, companies or scientific laboratories, and are based on real-life problems. Participants submit the results of their classification or prediction algorithms, which are then put into competition via the website. The challenges are part of Prof.Stéphane Mallat's courseat the Collège de France ( ), and are also offered in numerous data science lectures in France and throughout the French-speaking world.

The 2023 edition is now launched with thirteen new challenges, organized in partnership with theÉcole Normale Supérieure and the Institut Louis Bachelier. The seminar presents the seven other data challenges for 2023 :

• Predicting end-of-session returns on the US equity market (CFM) ;
• Robustness to distribution change and ambiguity (EffiSciences) ;
• The curse of the table of contents (l'Autorité des Marchés Financiers) ;
• Real-time prediction of train occupancy (SNCF-Transilien) ;
• Reinforcement learning for buildings' carbon footprint (Accenta) ;
• Learning radiological anatomy using few shots learning (Raidium) ;
• Colloids : how big can we get ? (ESPCI Paris).

Detailed presentations of the 2023 data challenges are available on the Challenge datawebsite .