Xue Wang

—

Unclaimed · AI-structured profileOpenAlex

Early-career machine learning researcher on tabular foundation models and in-context learning mechanisms

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-09-28) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Not yet scored: this OpenAlex record has too thin a publication history for an output measure. The axis would track cumulative work counts, per-year production, and output volume relative to field norms.

Impact—

Not yet scored: citation signals are not yet meaningful for this early record, so impact cannot be assessed. The axis would measure citations to the 2026 tabular-ICL papers and uptake beyond the immediate team.

Focus—

Not yet scored: OpenAlex topic tags spread across five areas with a count of one each, showing breadth without a settled concentration. The axis would measure how consistently works cluster within a primary research area.

Collaboration46/100

Scores 46 of 100. The record lists 9 distinct coauthors, with recurring collaborators including Tian Zhou, Beverly Jin, Wenwei Wang, Bingqing Peng, Mengni Ye, Jinjie Gu, Liang Sun, and Linxiao Yang. On both 2026 catalog papers Xue Wang appears in a middle authorship position rather than first or corresponding, consistent with a collaborative contributor inside a larger tabular-foundation-model team rather than a solo line of work.

Activity32/100

Scores 32 of 100. OpenAlex records 2 works in the recent three-year window, both surfacing in 2026 per the per-year counts — the two arXiv tabular-foundation-model papers reviewed in the Evolve catalog. Activity here reflects a concentrated 2026 burst on tabular in-context learning rather than sustained multi-year production, so the score should be read as an early-career starting point, not an established trajectory.

Reproducibility—

Not yet scored: no aggregate reproducibility signal exists for this record. Evolve does rate both 2026 catalog papers as medium readiness; the axis itself would measure code, data, and benchmark availability across all works.

Research summary

Xue Wang is an early-stage machine learning researcher whose public record, resolved from OpenAlex bibliographic data, centers on tabular foundation models and how they compute in context. The record is young: OpenAlex lists an h-index and i10-index of 0, only 2 works in the recent three-year window, and both surface in 2026. Evolve's catalog contains two 2026 arXiv papers with Xue Wang in a middle authorship position, neither as corresponding author. The first, Support-Compiled Feature Folding, folds wide tables into bounded leaves for tabular in-context learning, reporting accuracy and NLL gains across six backbones with 2.1–2.4x memory savings, with benefits hinging on feature selection. The second asks what tabular foundation models compute in context and derives an FFN-free, attention-gated model that beats TabPFN-3 by 31.4 Elo on TabArena, with strong intervention evidence although headline gains are partly post-hoc. OpenAlex topic tags spread across five areas — neural network applications, information retrieval and search behavior, graph neural networks, domain adaptation and few-shot learning, and multimodal machine learning — each with a count of one, so no stable focus has consolidated yet. Radar scores show collaboration at 46 with 9 distinct coauthors, including recurring collaborators such as Tian Zhou, Beverly Jin, and Wenwei Wang, and activity at 32, reflecting a concentrated 2026 start rather than an established publication record. Both papers carry medium reproducibility readiness.

Collaboration pattern

Both catalog papers list Xue Wang in a middle authorship position with no corresponding role, alongside a pool of 9 distinct OpenAlex coauthors. Recurring names include Tian Zhou, Beverly Jin, Wenwei Wang, Bingqing Peng, Mengni Ye, Jinjie Gu, Liang Sun, and Linxiao Yang, suggesting a stable tabular-foundation-model group; Linxiao Yang appears twice in the coauthor list. The collaboration score of 46 reflects moderate co-authorship intensity for such an early record.

Research keywords

tabular foundation modelstabular deep learningin-context learningrepresentation refinementattention gatingFFN-free transformerTabPFNmemory-efficient inferencefeature selectionTabArena benchmarking

Highlights

Bounded-leaf folding cuts memory across tabular foundation models
On the Support-Compiled Feature Folding paper, Evolve's verdict reports accuracy and NLL gains across six backbones with 2.1–2.4x memory savings for tabular in-context learning, noting the benefits hinge on feature selection. Reproducibility readiness: medium.
An FFN-free, attention-gated account of tabular in-context learning
The representation-refinement paper derives an FFN-free, attention-gated tabular ICL model that beats TabPFN-3 by 31.4 Elo on TabArena; Evolve notes strong intervention evidence while flagging that headline gains are partly post-hoc. Reproducibility readiness: medium.
Consistent contributor role across both 2026 releases
On both 2026 arXiv papers Xue Wang appears in a middle authorship position without a corresponding role, within a coauthor pool of 9 — a back-to-back pair on tabular ICL efficiency and mechanism within a single year.

Timeline

2026OpenAlex records 2 works for this author in 2026 — the entirety of the visible publication record to date.
2026Co-authored (middle position) 'Support-Compiled Feature Folding' on arXiv, reporting accuracy/NLL gains across six backbones with 2.1–2.4x memory savings for tabular in-context learning.
2026Co-authored (middle position) the in-situ representation refinement study on arXiv, deriving an FFN-free, attention-gated tabular ICL model that beats TabPFN-3 by 31.4 Elo on TabArena.

Papers in the Evolve catalog

Frequent co-authors

Tian Zhou×2 Beverly Jin×2 Wenwei Wang×2 Bingqing Peng×2 Mengni Ye×2 Jinjie Gu×2 Liang Sun×2 Linxiao Yang×1 Linxiao Yang×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometrics come from OpenAlex only; this record carries no institution, country, or ORCID data.
h-index and i10-index are 0 and only 2 recent three-year works are recorded, so output and impact indicators are not meaningful at this stage.
'Xue Wang' is a common name; this profile is name-resolved from public bibliographic data and may merge or omit unrelated namesakes.
Coverage is limited to the two 2026 arXiv papers reviewed by Evolve; any other works fall outside this profile's scope.