Evolve.Pub
Sign inScan your paper
BP

Bingqing Peng

—

Unclaimed · AI-structured profileOpenAlex

ML researcher on tabular foundation models, in-context representation learning, and glucose-based health AI

3

papers here

0

reviewed by Evolve

0

first-author

0

corresponding

—

works · OpenAlex

0

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: insufficient dataOutputImpact: insufficient dataImpactFocus: insufficient dataFocusCollaboration: 48/100CollaborationActivity: 40/100ActivityReproducibility: insufficient dataReproducibility

Output—

Output cannot be scored yet: OpenAlex records an h-index and i10-index of 0 and the Evolve catalog holds only three 2026 preprints. This axis would measure publication volume, venue quality, and authorship role as the record matures.

Impact—

Impact data is insufficient: recorded citation indicators are 0 and the 2026 preprints are too recent for uptake. This axis would measure citations, usage, and influence on follow-up work once citations accrue.

Focus—

Topical focus cannot be scored: the five OpenAlex topics each map to a single work, offering no concentration signal. This axis would measure how consistently the research program clusters around a stable specialization.

Collaboration48/100

Bingqing Peng works in team settings: the record shows ten distinct coauthors across three works, with frequent collaborators including Tian Zhou, Wenwei Wang, Mengni Ye, Beverly Jin, Jinjie Gu, Liang Sun, Linxiao Yang, Xue Wang, and Zuyi Zhu. All three catalog papers carry a middle-author position and none is corresponding, placing the collaboration score at a moderate 48 — an established contributor within groups rather than a listed lead.

Activity40/100

Research activity is recent and concentrated: OpenAlex counts three works in the last three years, all appearing in 2026, which sets the activity score at 40. The record shows no earlier publication years, so current output looks like the start of a publishing cadence rather than an established multi-year rate; activity should be re-read once a second year enters the counts-by-year series.

Reproducibility—

Reproducibility is unscored: Evolve provides no reproducibility-readiness assessment for these papers, and code or data availability is not recorded. This axis would measure whether artifacts, implementations, and datasets are released and documented.

Research summary

Bingqing Peng is an early-career machine learning researcher whose recent work centers on tabular foundation models and their emerging applications to health data. The Evolve catalog lists three 2026 arXiv preprints, each with a middle-author position: a memory-efficient feature-folding method for aggregating more evidence across tabular foundation models, a transferable evidence-reconstruction approach for longitudinal glucose representations, and a mechanistic study of what tabular foundation models compute in context through attention-gated representation refinement. OpenAlex topic assignments place the work across advanced neural network applications, graph neural networks, information retrieval and search behavior, diabetes management, and domain adaptation with few-shot learning — a breadth that is real but too thin (one work per topic) to establish a stable specialization. Bibliometric indicators are sparse: h-index and i10-index are recorded as 0, with three works in the last three years and ten distinct coauthors, yielding modest activity (40) and collaboration (48) scores on the Evolve radar, while output, impact, focus, and reproducibility lack sufficient data to score. The profile is name-resolved from public bibliographic data and should be read as an early-stage record that will firm up as these preprints accrue review, citations, and code releases.

Collaboration pattern

Bingqing Peng publishes in team settings: ten distinct coauthors appear across the three recent works, with recurring collaborators including Tian Zhou, Wenwei Wang, Mengni Ye, Beverly Jin, Jinjie Gu, Liang Sun, Linxiao Yang, Xue Wang, and Zuyi Zhu. Every catalog paper lists a middle-author position with no corresponding role, so current evidence points to a consistent team contributor rather than a group lead; the Evolve collaboration score is 48.

Research keywords

Tabular foundation modelsIn-context learningRepresentation learningAttention mechanismsMemory-efficient inferenceFeature foldingLongitudinal glucose modelingDiabetes managementDomain adaptationFew-shot learningTransfer learningGraph neural networks

Highlights

  • Memory-efficient evidence aggregation for tabular foundation models

    In the 2026 arXiv preprint 'Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models,' Peng contributed as a middle author to a method that folds features so more supporting evidence fits within tighter memory budgets across tabular foundation models.

  • Transferable representations for longitudinal glucose data

    The 2026 preprint 'Transferable Evidence Reconstruction for Longitudinal Glucose Representations' (middle-author contribution) reconstructs evidence from longitudinal glucose signals so representations transfer across contexts, connecting sequence modeling to diabetes management — an OpenAlex-listed topic for this record.

  • Interpreting what tabular foundation models compute in context

    In the 2026 preprint 'What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates,' Peng (middle author) helped examine how in-context representations are refined through attention-gated updates, an interpretability angle on tabular in-context learning.

Timeline

  1. 2026Co-authored (middle author) the arXiv preprint 'Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models' on memory-efficient evidence aggregation for tabular foundation models.
  2. 2026Co-authored the arXiv preprint 'Transferable Evidence Reconstruction for Longitudinal Glucose Representations,' linking representation learning with longitudinal glucose signals and diabetes management.
  3. 2026Co-authored the arXiv preprint 'What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates,' an interpretability study of in-context computation.

Papers in the Evolve catalog

Frequent co-authors

Tian Zhou×3Wenwei Wang×3Mengni Ye×3Beverly Jin×3Jinjie Gu×3Liang Sun×3Linxiao Yang×2Xue Wang×2Zuyi Zhu×1Linxiao Yang×1

Is this you? Claim this profile

Enter the email you published with. We'll send a one-time claim link; after claiming you can edit the introduction or hide this page. Requests from a mismatched address go to manual review.

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.

  • Bibliometric indicators come from OpenAlex only; h-index and i10-index are recorded as 0, reflecting a sparse early record rather than verified absence of impact.
  • All three catalog papers are 2026 arXiv preprints; peer-review status and final venues are not yet known.
  • No institution, country, or ORCID is available; 'Bingqing Peng' is a common name and this profile is name-resolved from public bibliographic data, so attribution may overlap with other researchers.
  • Evolve review metadata (verdict, keywords, reproducibility readiness) is not yet available for any of the three papers.
Bingqing Peng — Researcher Profile | Evolve.Pub