ML researcher in physiological time-series representation learning and tabular foundation-model in-context learning
2
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—
No output score is available: the OpenAlex record holds too few indexed works (2 recent works) to measure career-wide productivity. This axis would aggregate total publications, venue types, and yearly output across the full career.
Impact—
Impact data is insufficient for scoring: the indexed record reports h-index 0 and i10-index 0, reflecting a sparse OpenAlex profile rather than measured citation impact. This axis would track citations, h-index, and influence on follow-up work.
Focus—
No focus score: the snapshot lists 5 OpenAlex topics, each tied to a single work — spanning information retrieval, glucose monitoring, and graph neural networks — too thin to quantify topical concentration. This axis would measure thematic consistency over time.
Collaboration46/100
Scores 46/100 for collaboration, computed from an OpenAlex co-authorship record listing 9 distinct co-authors. Both cataloged 2026 arXiv papers place Linxiao Yang in a middle authorship position rather than first or corresponding author, consistent with contributing inside multi-person teams; named frequent co-authors include Tian Zhou, Bingqing Peng, Wenwei Wang, and Jinjie Gu.
Activity32/100
Scores 32/100 for activity. The OpenAlex record shows 2 works in the last three years, both indexed in 2026, with no indexed output in earlier years of the snapshot. This thin recent volume places current productivity low relative to established researchers, consistent with an early-stage profile just entering the indexed literature.
Reproducibility—
No reproducibility score exists, since artifact-level evidence is not aggregated for this profile. The axis would measure released code, data, and pretrained checkpoints. For context, Evolve's reviews rate both cataloged 2026 papers at medium reproducibility readiness.
Research summary
Linxiao Yang is an early-stage machine learning researcher whose indexed record and Evolve-reviewed output center on two 2026 arXiv preprints: one on self-supervised representation learning for continuous glucose monitoring data, and one on the internal mechanics of tabular foundation models. The first, 'Transferable Evidence Reconstruction for Longitudinal Glucose Representations,' proposes cross-group reader-transfer SSL for CGM signals, combining differentiable ridge regression and episodic meta-learning; Evolve's review credits strong, controlled gains across 14 tasks while flagging private pretraining data and benchmark-tuned development. The second, 'What Do Tabular Foundation Models Compute In Context?', derives an FFN-free, attention-gated in-context learning model reported to beat TabPFN-3 by 31.4 Elo on TabArena, with strong intervention evidence though partly post-hoc headline gains. In both papers Yang holds a middle authorship position, not corresponding author, within a nine-person co-author network (Tian Zhou, Bingqing Peng, Wenwei Wang, and others). Bibliometrically the profile is thin: OpenAlex lists 2 works in the last three years (both 2026), h-index and i10-index of 0, and topics spanning information retrieval, diabetes management, graph neural networks, glycemic control, and multimodal machine learning. Radar scores are collaboration 46/100 and activity 32/100, with output, impact, focus, and reproducibility unscored for lack of data — an emerging profile best read through paper-level review evidence rather than citation statistics.
Collaboration pattern
Collaboration scores 46/100 from an OpenAlex record with 9 distinct co-authors, matching the frequent-coauthor list: Tian Zhou, Bingqing Peng, Wenwei Wang, Mengni Ye, Beverly Jin, Jinjie Gu, Liang Sun, Zuyi Zhu, and Xue Wang. Both 2026 papers place Yang in a middle, non-corresponding position, indicating contribution within larger teams rather than solo or lead-author work; the co-author pool spans two distinct subfields, healthcare time-series ML and tabular foundation models.
Research keywords
Highlights
FFN-free, attention-gated account of tabular in-context learning
Cross-group reader-transfer SSL for continuous glucose monitoring
Timeline
- 2026arXiv preprint 'What Do Tabular Foundation Models Compute In Context?' appears — an FFN-free, attention-gated tabular ICL model reported at +31.4 Elo over TabPFN-3 on TabArena (middle authorship, not corresponding).
- 2026arXiv preprint 'Transferable Evidence Reconstruction for Longitudinal Glucose Representations' appears — cross-group reader-transfer SSL for CGM with controlled gains across 14 tasks (middle authorship, not corresponding).
- 2026OpenAlex snapshot indexes 2 works this year involving 9 distinct co-authors; h-index and i10-index stand at 0, marking an early-stage record.
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 figures come from OpenAlex only; identity is name-resolved from public bibliographic data, and 'Linxiao Yang' may correspond to more than one researcher.
- Institution, country, and ORCID are absent from the source data, so affiliation context is not represented.
- Citation indicators (h-index 0, i10-index 0) reflect a sparse indexed record covering 2 recent works, not a verified full publication history.
- Both cataloged papers are 2026 arXiv preprints reviewed by Evolve; no peer-reviewed venue or funding information is included.