BWhat changed
Applied Transformer layers to categorical-column embeddings, combined contextual outputs with continuous features in an MLP and studied masked/replaced-feature pretraining when labeled data are scarce.
Scientific scope
Applied Transformer layers to categorical-column embeddings, combined contextual outputs with continuous features in an MLP and studied masked/replaced-feature pretraining when labeled data are scarce. Original continuous inputs bypass the categorical attention stack, unlike later all-feature tokenizers. Missing/noisy-feature robustness and mean-AUC gains are limited to studied datasets/settings, not causal interpretation or universal superiority. General tabular architecture only; no biomedical dataset outcome is added.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Original continuous inputs bypass the categorical attention stack, unlike later all-feature tokenizers. Missing/noisy-feature robustness and mean-AUC gains are limited to studied datasets/settings, not causal interpretation or universal superiority. General tabular architecture only; no biomedical dataset outcome is added.
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.
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