lnu.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Q-Tab: Quantized Tabular Data Generator
University of Augsburg, Germany.
University of Augsburg, Germany.
Linnaeus University, School of Business and Economics, Department of Economics and Statistics. University of Augsburg, Germany.ORCID iD: 0000-0003-4704-5233
Munich University of Applied Sciences, Germany.
2026 (English)In: Proceedings of the 43rd International Conference on Machine Learning (ICML 2026): Seoul, South Korea. PMLR 306, 2026 / [ed] Tong Zhang; Miroslav Dudik; Martin Jaggi; Alekh Agarwal; Sharon Li; Dale Schuurmans; Jerry Zhu; Felix Berkenkamp; Hanze Dong; Alberto Bietti, ML Research Press , 2026, p. 137881-137901Conference paper, Published paper (Refereed)
Abstract [en]

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.

Place, publisher, year, edition, pages
ML Research Press , 2026. p. 137881-137901
Series
Proceedings of Machine Learning Research, ISSN 2640-3498 ; 306
National Category
Artificial Intelligence Probability Theory and Statistics
Research subject
Statistics/Econometrics
Identifiers
URN: urn:nbn:se:lnu:diva-150343OAI: oai:DiVA.org:lnu-150343DiVA, id: diva2:2113025
Conference
International Conference on Machine Learning, Seoul, South Korea, July 6-11, 2026
Available from: 2026-10-27 Created: 2026-10-27 Last updated: 2026-10-29Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Abstract

Authority records

Okhrin, Yarema

Search in DiVA

By author/editor
Okhrin, Yarema
By organisation
Department of Economics and Statistics
Artificial IntelligenceProbability Theory and Statistics

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 4 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf