Machine learning researcher working on tabular foundation models, representation learning, and AI for longitudinal health data
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—
Not yet scored: the OpenAlex record shows three 2026 arXiv preprints with h-index 0 and i10-index 0, too little published output to characterize productivity. This axis would measure total works, publication venues, and output trends over time.
Impact—
Not yet scored: citation indicators are zero in the raw record (h-index 0, i10-index 0), and the three 2026 preprints are too recent for uptake to accumulate. This axis would measure accumulated citations, influential-scholar counts, and downstream use of the work.
Focus—
Not yet scored: five OpenAlex topics appear once each — neural network applications, information retrieval, diabetes management, graph neural networks, and domain adaptation with few-shot learning — so no dominant specialization has stabilized. This axis would measure topic concentration and consistency over time.
Collaboration48/100
The collaboration score of 48 reflects a moderately connected early profile: the OpenAlex record lists 10 distinct coauthors, including frequent collaborators Tian Zhou, Bingqing Peng, Wenwei Wang, Mengni Ye, and Beverly Jin. On all three 2026 catalog papers Gu holds a middle authorship position without corresponding-author status, consistent with contributing inside established teams rather than directing an independent group so far.
Activity40/100
The activity score of 40 rests on a thin but very recent record: 3 works in the last three years, with all three falling in 2026 according to the per-year counts, each an arXiv preprint. This suggests a burst of current output rather than sustained multi-year productivity, so the score stays moderate pending further years of data.
Reproducibility—
Not yet scored: no repro-readiness assessments exist for the three catalog papers (no Evolve verdicts or keywords are available). This axis would measure code release, dataset availability, and how readily reported results can be independently reproduced.
Research summary
Jinjie Gu is an early-stage machine learning researcher whose public record, name-resolved from OpenAlex bibliographic data, centers on tabular foundation models and representation learning for structured and longitudinal health data. The 2026 snapshot lists three works, all arXiv preprints, in each of which Gu appears as a middle rather than lead or corresponding author. Two of them probe what tabular foundation models do and how far they can be pushed: Support-Compiled Feature Folding targets more evidence at lower memory across tabular foundation models, while the in-context study analyzes internal computation through attention-gated, in-situ representation refinement. The third applies transferable evidence reconstruction to longitudinal glucose representations, tying the same modeling ideas to diabetes management. Five OpenAlex topics — advanced neural network applications, information retrieval and search behavior, diabetes management and research, advanced graph neural networks, and domain adaptation with few-shot learning — each appear once, signaling breadth rather than a settled specialization. With h-index 0, i10-index 0, three works in the last three years, and ten distinct coauthors, the profile reads as the beginning of a research career with collaboration as its strongest measured dimension.
Collaboration pattern
Gu's record lists 10 distinct coauthors, with Tian Zhou, Bingqing Peng, Wenwei Wang, Mengni Ye, Beverly Jin, Liang Sun, Linxiao Yang, Xue Wang, and Zuyi Zhu appearing most often. Across all three 2026 catalog papers Gu is a middle author without corresponding status, consistent with contributing inside established teams. The same collaborators recur across tabular-model and glucose-representation projects, suggesting a stable, group-based working pattern rather than one-off or solo efforts.
Research keywords
Highlights
Pushing evidence density in tabular foundation models under memory limits
The 2026 arXiv preprint Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models proposes folding support-compiled features so models carry more evidence at lower memory across tabular foundation models; Gu contributes as a middle author, and no Evolve verdict or repro-readiness assessment exists yet.
Opening up the in-context computation of tabular foundation models
In the 2026 arXiv preprint What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates, the authors characterize internal computation as in-situ representation refinement through attention-gated updates; Gu is a middle author on this interpretability-oriented study.
Transferable evidence reconstruction for longitudinal glucose data
The 2026 arXiv preprint Transferable Evidence Reconstruction for Longitudinal Glucose Representations extends the group's representation-learning line to longitudinal glucose signals, connecting it to diabetes management; OpenAlex indexes a matching Diabetes Management and Research topic for Gu. Gu is a middle author.
Timeline
- 2026Appears in the Evolve catalog as a middle author on the arXiv preprint Support-Compiled Feature Folding: More Evidence at Lower Memory Across Tabular Foundation Models.
- 2026Co-authors the arXiv preprint Transferable Evidence Reconstruction for Longitudinal Glucose Representations, extending the work toward glucose and diabetes applications.
- 2026Co-authors the arXiv preprint What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates.
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.
- Bibliometrics come from an OpenAlex author record only (identity: openalex); no institution, country, or ORCID is attached, and the profile is name-resolved from public bibliographic data.
- Citation indicators are zero (h-index 0, i10-index 0) and all three papers are 2026 arXiv preprints, so the output, impact, focus, and reproducibility axes remain unscored.
- Gu is a middle, non-corresponding author on every cataloged paper; no first-author or corresponding-author work appears in the available data.
- No Evolve review verdicts, field labels, or repro-readiness assessments exist for the three papers, and no venue information beyond arXiv is available.