Challenges and Opportunities of Table Representation Learning (Dagstuhl Seminar 25182)

Download PDF
Open Webpage

Carsten Binnig, Julian Martin Eisenschlos, Madelon Hulsebos, Frank Hutter

Published: 2025, Last Modified: 05 Jan 2026Dagstuhl Reports 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: The growing volume and importance of structured data have sparked increasing interest in Table Representation Learning (TRL), an emerging field that leverages neural models to learn abstract, general-purpose representations for tabular data to support a wide range of downstream tasks such as tabular prediction, table question answering, tabular data cleaning, and many more. This seminar gathered the different communities (ML, NLP, IR, DB) who work on this topic to discuss the challenges & long-term vision of this field. From the organizers: Carsten Binnig, Julian Eisenschlos, Madelon Hulsebos, Frank Hutter.