Liang Sun

—

Unclaimed · AI-structured profileOpenAlex

Last author on 2026 work in tabular foundation models, in-context representation learning, and glucose-based health AI

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—

Output metrics are not yet computable: the OpenAlex record shows only three 2026 works with h-index 0, so this axis would normally measure publication volume and productivity trends once a longer publication history accumulates.

Impact—

Impact cannot be scored yet: h-index and i10-index both stand at 0 in the source data. This axis would measure citations, citation trajectories, and scholarly uptake once the 2026 preprints accumulate readership.

Focus—

Focus concentration is not scored: OpenAlex assigns five distinct topics, each with a single paper, so no dominant field has emerged. The axis would measure topical concentration once further works clarify the research agenda.

Collaboration48/100

Collaboration scores 48 on the radar. The OpenAlex record lists 10 distinct coauthors, and the recurring collaborator list — Tian Zhou, Bingqing Peng, Wenwei Wang, Mengni Ye, Beverly Jin, Jinjie Gu, Linxiao Yang, Xue Wang, and Zuyi Zhu — points to a moderately sized, rotating network. Holding the last-author position on all three 2026 preprints suggests a senior, supervisory role within these teams, though the score remains mid-range.

Activity40/100

Activity scores 40. The counts-by-year series records 3 works in 2026 and none in earlier years, matching recent3yWorks = 3. All three outputs are arXiv preprints posted within a single year, indicating a concentrated burst of new work; with such a short record, the activity score reflects a brief window rather than sustained multi-year productivity.

Reproducibility—

Reproducibility is not yet assessed: no Evolve verdict or repro-readiness rating exists for the three cataloged preprints. The axis would measure code release, data availability, and artifact quality once the papers are formally reviewed.

Research summary

Liang Sun is an early-stage machine learning researcher whose profile — name-resolved from public OpenAlex bibliographic data — currently rests on three 2026 arXiv preprints, each with Liang Sun in the last-author position. The work follows two connected threads. The first is tabular foundation models: one preprint introduces support-compiled feature folding to carry more evidence at lower memory across tabular foundation models, while another asks what these models compute in context, characterizing in-situ representation refinement through attention-gated updates. The second thread applies representation learning to physiological data, reconstructing transferable evidence for longitudinal glucose representations — a direction relevant to diabetes management. OpenAlex topic assignments mirror this spread, listing advanced neural network applications, graph neural networks, domain adaptation and few-shot learning, information retrieval and search behavior, and diabetes management, one paper each. Citation indicators remain at zero (h-index 0, i10-index 0), consistent with output that begins and ends in 2026: three works in the last three years and ten distinct coauthors. No affiliation, country, or career history is present in the source data.

Collaboration pattern

The record lists 10 distinct coauthors within a very short window, with collaboration scoring 48 on the radar. Tian Zhou, Bingqing Peng, Wenwei Wang, Mengni Ye, Beverly Jin, Jinjie Gu, Linxiao Yang, Xue Wang, and Zuyi Zhu recur as frequent collaborators, suggesting stable small teams rather than one large consortium. Liang Sun occupies the last-author seat on all three 2026 preprints, a position typically associated with supervising the research.

Research keywords

Tabular foundation modelsIn-context learningRepresentation refinementAttention mechanismsDomain adaptationFew-shot learningGraph neural networksLongitudinal glucose modelingDiabetes managementMemory-efficient inferenceInformation retrieval

Highlights

Support-compiled feature folding for tabular foundation models
A 2026 arXiv preprint with Liang Sun as last author proposing support-compiled feature folding, aiming to carry more evidence at lower memory across tabular foundation models.
In-situ representation refinement inside tabular models
The 2026 study 'What Do Tabular Foundation Models Compute In Context?' examines in-context computation and describes representation refinement through attention-gated updates.
Transferable evidence for longitudinal glucose representations
A 2026 preprint on transferable evidence reconstruction for longitudinal glucose representations, extending this representation-learning line of work toward diabetes-related physiological data.

Timeline

2026The OpenAlex record begins: three arXiv preprints appear in 2026, the only year with recorded output (recent3yWorks = 3).
2026Support-Compiled Feature Folding, on more evidence at lower memory across tabular foundation models, is posted with Liang Sun as last author.
2026Transferable Evidence Reconstruction for Longitudinal Glucose Representations is posted, connecting the agenda to diabetes management.
2026The in-context study of tabular foundation models, describing attention-gated in-situ representation refinement, is posted.

Papers in the Evolve catalog

Frequent co-authors

Tian Zhou×3 Bingqing Peng×3 Wenwei Wang×3 Mengni Ye×3 Beverly Jin×3 Jinjie Gu×3 Linxiao Yang×2 Xue Wang×2 Zuyi Zhu×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.

All bibliometrics come from a single OpenAlex record (identity: openalex); no other citation databases were consulted.
Name resolution: 'Liang Sun' is a common name; the profile is built from public bibliographic data, with institution, country, and ORCID absent from the source.
The three cataloged papers are arXiv preprints with no Evolve verdict, field tags, or repro-readiness ratings yet.
h-index and i10-index are reported as 0 and all output sits in 2026, so any impact-related conclusion would be premature.