Statistical learning theorist at Zhejiang University: kernel methods, distributed learning, target-shift regression
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
15
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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: this axis measures total research output and its growth over time, including annual publication counts and authorship volume. The underlying cross-source production data is insufficient in the current snapshot, so no estimate is offered.
Impact—
Not yet scored: this axis measures citation-based influence, such as accumulated citations, h-index trajectory, and field-normalized impact. The current snapshot lacks the validated citation data needed for this axis, so no score or estimate is provided.
Focus0/100
The focus score is recorded as 0, but the underlying field distribution suggests this reflects label fragmentation rather than real breadth: the record splits evenly (0.5/0.5) between "statistical learning theory" and its Chinese-language equivalent covering kernel methods and distributed learning — labels denoting the same single field. Topic counts likewise cluster tightly around gradient-based optimization and kernel learning. Interpret this axis with care.
Collaboration28/100
The collaboration score of 28 is low, driven by a coauthor pool of only three distinct collaborators in the indexed data. Repeated co-authorship with Jun-Yi Meng, Yuan Mao, and Ren-Rui Liu indicates a stable, small working group. On the two reviewed 2026 preprints Guo appears as last author on one and a middle author on the other, neither in the corresponding role — a senior contributor position within a compact team.
Activity79/100
The activity score of 79 reflects strong recent productivity: fourteen indexed works in the last three years, with four to five outputs per year from 2024 through 2026 in the counts-by-year series. This continues an active record reaching back to 2011 with a six-work peak in 2017, though the series also shows thin years (a single work in 2012, 2014, 2015, 2016, 2018) and a gap across 2020–2021.
Reproducibility38/100
The reproducibility score of 38 is the mean readiness across two assessed papers (reproPapers: 2, reproMean: 38). Both are purely theoretical: the density-ratio-estimation preprint is rated low (no experiments; optimality asserted rather than fully proved), while the distributed kernel gradient descent analysis is rated medium. Rigor is high, but the absence of empirical artifacts means results cannot currently be reproduced from code or data.
Research summary
Zheng-Chu Guo is a researcher at Zhejiang University working in statistical learning theory, with particular depth in kernel methods and distributed learning. The OpenAlex-indexed record shows sustained productivity since 2011, including a six-work peak in 2017 and strong recent output — fourteen works in the last three years, with four to five indexed outputs each year from 2024 through 2026. Citation indicators include an h-index of 15 and an i10-index of 17. Topic assignment concentrates on stochastic gradient optimization techniques (31 works), machine learning and algorithms (27), machine learning and ELM (14), statistical methods and inference (10), and domain adaptation with few-shot learning (8). Two 2026 arXiv preprints reviewed by Evolve — a finite-sample treatment of density ratio estimation under target shift, and a generalization analysis of distributed kernel-based robust gradient descent — are both purely theoretical: they deliver minimax-rate and generalization guarantees without experiments, and carry low-to-medium reproducibility readiness. The coauthor pool is small (three distinct collaborators), pointing to a compact, stable working group rather than a broad collaborative network.
Collaboration pattern
Collaboration is narrow and stable: only three distinct coauthors appear in the indexed record, and Jun-Yi Meng, Yuan Mao, and Ren-Rui Liu recur as the frequent partners. The collaboration score of 28 reflects this compactness rather than inactivity — output remains steady. On the two reviewed 2026 preprints, Guo takes the last authorship on one and a middle position on the other, and is corresponding author on neither, consistent with a senior contributor within a small, theory-focused group.
Research keywords
Highlights
Minimax-optimal density ratio estimation under target shift
Optimal rates for distributed kernel-based robust gradient descent
A purely theoretical program with an empirical gap
Across both reviewed papers the pattern is consistent: sharp minimax and generalization guarantees with no accompanying experiments. Closing this gap — releasing code or numerical checks for the target-shift and distributed settings — is the clearest route to lifting the current reproducibility mean of 38.
Timeline
- 2011Earliest indexed outputs appear in the OpenAlex record, with four works.
- 2017Most productive year in the record: six indexed works.
- 2019Three indexed works; the last active year before a two-year gap.
- 2022Activity resumes after the 2020–2021 gap, with four indexed works.
- 2025Five indexed works, matching the following year's pace.
- 2026Two arXiv preprints enter the Evolve catalog: density ratio estimation under target shift (last author) and distributed kernel-based robust gradient descent (middle author).
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 (h-index 15, i10-index 17, counts by year, topics) come from an OpenAlex-indexed snapshot; identity is name-resolved from public bibliographic data, so name collisions are possible.
- Output and impact dimensions are unscored because the underlying radar data for those axes is missing, not because the axes lack relevance.
- Only two papers have been fully reviewed by Evolve, both 2026 arXiv preprints; the reproducibility mean of 38 rests on those two assessments alone.
- The focus score of 0 appears to reflect bilingual field-label splitting (one field recorded separately in Chinese and English) rather than genuine topical dispersion.