Why Deep Learning Failed on Tables for a Decade - Frank Hutter @MachineLearningStreetTalk
Why Deep Learning Failed on Tables for a Decade - Frank Hutter  @MachineLearningStreetTalk
Uploaded September 2026 | Updated November 2026, 3 days ago
Frank Hutter, co-founder of Prior Labs, on why deep learning struggled with tabular data for a decade: tables are messy and heterogeneous, hyped models like TabNet did not generalise to new datasets, and there was no ImageNet of tables. The breakthrough came from learning to transfer at the level of patterns across many different tables; TabPFN-3.5 now tops the TabArena benchmark.

Full interview on MLST: youtube.com/watch?v=72Im-Mm5JKs

References:
TabNet (Arik & Pfister): arxiv.org/abs/1908.07442
TabPFN-3.5 Technical Report: arxiv.org/abs/2609.17895

#Shorts
Why Deep Learning Failed on Tables for a Decade - Frank HutterDon’t use NNs for simulation (Johannes Brandstetter)AlphaGo’s Move 37 Wasn’t Creative? | Edward HughesThe Trojan Horse in AI-Written MathematicsHow a Robot Tests an Action Before It MovesWild breakthrough on Math after 56 years... [Exclusive]Learning at test time in LLMs [Jonas Hübotter]AI is just the beginning (Anil Ananthaswamy)Why Scientists Cant Rebuild a Polaroid Camera [César Hidalgo]François Chollet on OpenAI o-models and ARCIs AI Just a Library? Prof. Krakauer Explains the ConfusionIt Was Fred That Broke the Window (Michael Tomasello)
Machine Learning Street Talk |

Why Deep Learning Failed on Tables for a Decade - Frank Hutter

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER