ML researcher on tabular foundation models, in-context representation refinement, and health applications
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 data for scoring: this axis measures cumulative publication volume and productivity trends over time. Beyond the 3 recent three-year works, the snapshot provides no full works count or multi-year history, so output cannot be scored yet.
Impact—
Insufficient data for scoring: this axis measures citation-based influence such as h-index, i10-index, and citation counts. The snapshot records h-index 0 and i10-index 0, which cannot yet distinguish low influence from a newly indexed profile.
Focus—
Insufficient data for scoring: this axis measures topical concentration versus dispersion. Five OpenAlex topics appear once each in the snapshot, too few indexed works to establish a stable research focus, so no focus score is computed.
Collaboration48/100
The collaboration dimension scores 48 on a 0-100 scale, computed from a snapshot recording 10 distinct co-authors across the 3 works in the recent three-year window. The three cataloged papers all list a middle author position with no corresponding-author role, indicating participation in team-based research rather than independent leadership. Recurring names such as Tian Zhou, Bingqing Peng, and Wenwei Wang point to a stable core collaborator group, though no senior mentor or long-term partner can be confirmed from this data.
Activity40/100
The activity dimension scores 40 on a 0-100 scale. The snapshot's counts-by-year data contains a single entry — 3 works in 2026 — and the recent three-year works count is also 3, meaning every indexed output falls in one year with no recorded earlier publications. Because the record appears newly created, this score reflects a short observable window rather than a slowdown; it should rise or fall as 2026 and later years accumulate.
Reproducibility—
Insufficient data for scoring: this axis measures reproducibility readiness — released code, datasets, and replication materials. None of the 3 cataloged papers carries a reproducibility assessment in this snapshot, so no score can be assigned.
Research summary
Beverly Jin is a machine learning researcher whose recently indexed record centers on tabular foundation models and their emerging applications. The Evolve catalog lists three 2026 arXiv papers, in each of which the researcher holds a middle author position: 'Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models'; 'What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates'; and 'Transferable Evidence Reconstruction for Longitudinal Glucose Representations'. Together they span efficiency-oriented methods for tabular data, mechanistic analysis of how tabular transformers refine representations in context, and transferable modeling of longitudinal glucose signals, connecting the portfolio to diabetes management research. OpenAlex topic assignments mirror this breadth, with one work each counted for advanced neural network applications, information retrieval and search behavior, diabetes management and research, advanced graph neural networks, and domain adaptation and few-shot learning. The snapshot records 3 works in the recent three-year window, 10 distinct co-authors, a collaboration score of 48, and an activity score of 40; citation indicators stand at h-index 0 and i10-index 0, consistent with a newly indexed profile rather than an established citation record. This is an early-stage profile: the output, impact, focus, and reproducibility dimensions lack sufficient data for scoring and should be reassessed as the 2026 record accumulates.
Collaboration pattern
The snapshot records 10 distinct co-authors across 3 recent works, yielding a collaboration score of 48. All cataloged papers place the researcher in a middle author position without corresponding-author status, consistent with contributing within larger teams. Recurring collaborator names include Tian Zhou, Bingqing Peng, Wenwei Wang, Mengni Ye, Jinjie Gu, and Liang Sun, while Linxiao Yang appears twice in the list, suggesting at least one repeated collaboration. Independent or leading-author work is not yet visible in this record.
Research keywords
Highlights
Efficiency-oriented work on tabular foundation models
The 2026 arXiv paper 'Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models' pursues holding more evidence at lower memory across tabular foundation models — the clearest efficiency-driven contribution among the three cataloged works.
Interpreting what tabular transformers compute in context
'What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates' (2026) probes how tabular foundation models refine representations in situ through attention-gated updates, adding a mechanistic perspective to the same model family.
Extending toward longitudinal glucose and diabetes research
'Transferable Evidence Reconstruction for Longitudinal Glucose Representations' (2026) builds transferable representations for longitudinal glucose signals, matching the OpenAlex topic 'Diabetes Management and Research' and connecting the tabular-models portfolio to health applications.
Breadth across topics rather than a settled specialty
OpenAlex assigns five topics with one work each — advanced neural network applications, information retrieval and search behavior, diabetes management and research, advanced graph neural networks, and domain adaptation and few-shot learning — indicating an exploratory, breadth-first stage.
Timeline
- 2026Co-authored 'Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models' on arXiv, in a middle author position.
- 2026Co-authored 'Transferable Evidence Reconstruction for Longitudinal Glucose Representations' on arXiv, in a middle author position.
- 2026Co-authored 'What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates' on arXiv, in a middle author position.
Papers in the Evolve catalog
Frequent co-authors
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 are derived from OpenAlex only; no ORCID or institutional affiliation is linked to this record.
- All three cataloged papers are 2026 arXiv preprints; peer-reviewed venues and Evolve review verdicts are not yet available.
- Citation indicators (h-index 0, i10-index 0) reflect a newly indexed profile and should not be read as a quality judgment.
- The name 'Beverly Jin' is resolved from public bibliographic data; without ORCID linkage, identity certainty is limited.