Q-Tab: Quantized Tabular Data Generator

Julian Wustl, Philipp Haid, Yarema Okhrin, Claudius Schnörr
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:137881-137901, 2026.

Abstract

Codebook-based generators built on masked language model (MLM) transformers have become highly effective in text and vision, yet remain underused for tabular data. This is because codebooks typically act as information bottlenecks, whereas synthetic tabular generation requires a code space larger than the training sample, with additional codes trained to support new tabular rows. We address this gap with Q-Tab, a codebook-based tabular generator that uses lookup-free quantization (LFQ) with residual corruption to jointly tokenize numerical variables, categorical variables are directly one-hot tokenized. A BERT-style MLM captures dependencies in the token space and can then be sampled from. Corruption propagates reconstruction supervision across the numerical code space, but under joint encoder–decoder training induces a moving-target regression problem whose difficulty depends on the corruption structure. This motivates residual LFQ as the quantization mechanism, balancing broader supervision with locality. Q-Tab achieves state-of-the-art predictive utility and label prediction, while matching the distributional fidelity of diffusion-based generators.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-wustl26a, title = {Q-Tab: Quantized Tabular Data Generator}, author = {Wustl, Julian and Haid, Philipp and Okhrin, Yarema and Schn\"{o}rr, Claudius}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {137881--137901}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/wustl26a/wustl26a.pdf}, url = {https://proceedings.mlr.press/v306/wustl26a.html}, abstract = {Codebook-based generators built on masked language model (MLM) transformers have become highly effective in text and vision, yet remain underused for tabular data. This is because codebooks typically act as information bottlenecks, whereas synthetic tabular generation requires a code space larger than the training sample, with additional codes trained to support new tabular rows. We address this gap with Q-Tab, a codebook-based tabular generator that uses lookup-free quantization (LFQ) with residual corruption to jointly tokenize numerical variables, categorical variables are directly one-hot tokenized. A BERT-style MLM captures dependencies in the token space and can then be sampled from. Corruption propagates reconstruction supervision across the numerical code space, but under joint encoder–decoder training induces a moving-target regression problem whose difficulty depends on the corruption structure. This motivates residual LFQ as the quantization mechanism, balancing broader supervision with locality. Q-Tab achieves state-of-the-art predictive utility and label prediction, while matching the distributional fidelity of diffusion-based generators.} }
Endnote
%0 Conference Paper %T Q-Tab: Quantized Tabular Data Generator %A Julian Wustl %A Philipp Haid %A Yarema Okhrin %A Claudius Schnörr %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-wustl26a %I PMLR %P 137881--137901 %U https://proceedings.mlr.press/v306/wustl26a.html %V 306 %X Codebook-based generators built on masked language model (MLM) transformers have become highly effective in text and vision, yet remain underused for tabular data. This is because codebooks typically act as information bottlenecks, whereas synthetic tabular generation requires a code space larger than the training sample, with additional codes trained to support new tabular rows. We address this gap with Q-Tab, a codebook-based tabular generator that uses lookup-free quantization (LFQ) with residual corruption to jointly tokenize numerical variables, categorical variables are directly one-hot tokenized. A BERT-style MLM captures dependencies in the token space and can then be sampled from. Corruption propagates reconstruction supervision across the numerical code space, but under joint encoder–decoder training induces a moving-target regression problem whose difficulty depends on the corruption structure. This motivates residual LFQ as the quantization mechanism, balancing broader supervision with locality. Q-Tab achieves state-of-the-art predictive utility and label prediction, while matching the distributional fidelity of diffusion-based generators.
APA
Wustl, J., Haid, P., Okhrin, Y. & Schnörr, C.. (2026). Q-Tab: Quantized Tabular Data Generator. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:137881-137901 Available from https://proceedings.mlr.press/v306/wustl26a.html.

Related Material