TabNet: sequential sparse feature selection for tabular prediction

What changed

Used input-dependent sparse attention masks to select features at successive decision steps, transformed selected features and aggregated their contributions for prediction and mask-based interpretation.

Scientific scope

Used input-dependent sparse attention masks to select features at successive decision steps, transformed selected features and aggregated their contributions for prediction and mask-based interpretation. Sparse masks and attribution visualizations do not prove causal explanations, unique feature importance or universal accuracy gains. Sparsemax and gated/residual building blocks are inherited. This original v1 architecture is not silently expanded with later unsupervised-pretraining variants or medical outcomes.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Sparse masks and attribution visualizations do not prove causal explanations, unique feature importance or universal accuracy gains. Sparsemax and gated/residual building blocks are inherited. This original v1 architecture is not silently expanded with later unsupervised-pretraining variants or medical outcomes.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• TabNet: Attentive Interpretable Tabular Learning

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction