AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Revisiting Deep Learning Models for Tabular Data

Yury Gorishniy, Ivan Rubachev, Valentin Khrulkov, Artem Babenko

NeurIPS 2021 · Architectures and efficiency › RNNs and normalization · approach: Dataset distillation and data-efficient training · 115 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add feature biases to Feature Tokenizer (ablation)FT-Transformer · Covertypeaccuracy: 0.964 -> 0.970Improved
add feature biases to Feature Tokenizer (ablation)FT-Transformer · YearRMSE: 8.843 -> 8.855Hurt
vanilla Transformer backbone with [CLS] and feature biases instead of AutoInt design (ablation)attention-based tabular model · Covertypeaccuracy: 0.934 -> 0.970Improved
FT-Transformer instead of ResNettabular DL · 11 tabular datasetsaverage rank: 3.3 (1.8) -> 1.8 (1.2)Improved
tuned GBDT ensembles instead of FT-Transformer ensemblestabular models · California HousingRMSE: 0.448 -> 0.423Improved
FT-Transformer instead of GBDTtabular models · Helenaaccuracy: 0.388 -> 0.398Improved
tune hyperparameters of simple baselines (MLP, ResNet)MLP/ResNet · 11 tabular datasetsrankImproved
ensemblingFT-Transformer · California HousingRMSE: 0.459 -> 0.448Improved
attention-map feature importance instead of Integrated GradientsFT-Transformer · Yearrank correlation with permutation test: 0.50 (0.03) -> 0.92 (0.01)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy