Wenhao Wang

National University of Defense Technology · China

Unclaimed · AI-structured profileOpenAlex

Early-career researcher at NUDT spanning RL theory and computer-use agents, with strong reproducibility

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

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—

Output cannot be scored yet: the record shows only 9 works in 2026, with no full publication history, so cumulative volume and first/corresponding-author share remain unmeasured. This axis would track works, authorship positions, and venues over time.

Impact—

Impact cannot be scored yet: available signals are limited to h-index 1 and i10-index 0, too sparse to characterize influence. This axis would measure citation counts, citing works, and field-normalized impact once citation data accumulates.

Focus0/100

Focus scores 0/100: the field distribution splits exactly 50/50 between 'artificial intelligence — computer-use agents' and 'reinforcement learning (machine learning theory)', and OpenAlex topics scatter across five areas — usability and UI design (3), privacy-preserving technologies (3), RL in robotics (2), AI planning (2), and web data mining (2). The record is spread rather than concentrated, consistent with an early-career profile still exploring directions.

Collaboration55/100

Collaboration scores 55/100. The record lists 15 distinct co-authors, and 12 recur frequently, including Haoting Shi, Weicheng Fang, Yaozhong Liang, Tian Jin, Siheng Chen, and Yanfeng Wang. On both reviewed papers Wang is a non-corresponding co-author — last position on the theory paper, middle on CUA-Universe — indicating participation within sizeable, stable teams rather than solo or leading roles.

Activity67/100

Activity scores 67/100. All 9 recorded works fall in 2026, matching recent3yWorks = 9, so the recent output rate is solid even though the record's h-index is only 1 and i10-index 0. The concentration in a single year suggests a fast-moving, recent publication phase rather than an established multi-year record.

Reproducibility78/100

Reproducibility scores 78/100, averaged over 2 reviewed papers. The last-authored emphatic TD paper rates high — it ships a rigorous counterexample and certified experiments, though only on tabular tasks. CUA-Universe rates medium: the pipeline and benchmark are described, but internal number conflicts and benchmark circularity temper confidence. The average masks real variance between the two.

Research summary

Wenhao Wang is a researcher at the National University of Defense Technology (China) whose profile is name-resolved from OpenAlex public bibliographic data. The record shows an early-stage career: h-index 1, i10-index 0, and 9 works recorded in 2026, alongside 15 distinct co-authors and 12 recurring collaborators. Reviewed work spans two distinct directions, each contributing half of the field distribution. In reinforcement learning theory, the last-authored 2026 paper on regularized emphatic temporal-difference learning stabilizes constant-stepsize off-policy TD through a leaky scalar state, backed by a rigorous counterexample and certified experiments on tabular tasks. In AI computer-use agents, a middle-authored 2026 paper presents CUA-Universe, a hybrid GUI+CLI agent environment covering 16 apps with a 9B model, reporting large gains on its own benchmark and OSWorld. Reproducibility is a visible strength: the two reviewed papers average 78/100, spanning high and medium readiness. Topic traces in usability and UI design, privacy-preserving technologies, RL robotics, AI planning, and web data mining suggest broader applied interests, though citation aggregates remain too sparse to score impact.

Collaboration pattern

The OpenAlex record lists 15 distinct co-authors, with 12 recurring collaborators including Haoting Shi, Weicheng Fang, Yaozhong Liang, Tian Jin, Siheng Chen, and Yanfeng Wang. Collaboration scores 55/100. On both reviewed papers Wang is a non-corresponding co-author — last author on the reinforcement-learning theory paper and middle author on CUA-Universe — consistent with contributing within larger, stable teams rather than leading.

Research keywords

reinforcement learning theoryoff-policy learningemphatic temporal-difference learningconstant-stepsize stabilitystochastic approximationLyapunov exponentscomputer-use agentsGUI agentsCLI agentshybrid interactionenvironment synthesistask synthesis

Highlights

Certified stability fix for emphatic TD under constant stepsizes
As last author on this 2026 arXiv paper, Wang contributes to regularized emphatic temporal-difference learning: a leaky scalar state resolves constant-stepsize instability, supported by a rigorous counterexample and certified experiments. Evolve notes the empirical scope is limited to tabular tasks, and reproducibility readiness is rated high.
CUA-Universe: hybrid GUI+CLI environment for computer-use agents
As a middle author, Wang contributed to a scalable, dynamic environment spanning 16 apps with a 9B model, reporting large gains on its own benchmark and OSWorld. Evolve flags internal number conflicts and benchmark circularity, and rates reproducibility readiness medium.
Reproducibility as a cross-paper strength
Two Evolve-reviewed papers average 78/100 on reproducibility, spanning high-readiness certified theory experiments and a medium-readiness agent environment — a spread reflecting careful experimental practice across both halves of Wang's two-field profile.

Timeline

20269 works appear in the OpenAlex record, all concentrated in this single year — the only year with recorded output.
2026Co-authors (middle position) the CUA-Universe paper on a scalable, dynamic environment for hybrid GUI+CLI agents.
2026Appears as last author on the regularized emphatic temporal-difference learning paper, with high reproducibility readiness.

Papers in the Evolve catalog

Frequent co-authors

Haoting Shi×1 Weicheng Fang×1 Yaozhong Liang×1 Tian Jin×1 Pengxiang Zhao×1 Guangyi Liu×1 Siheng Chen×1 Yanfeng Wang×1 Xingguo Chen×1 Zhaohui Wu×1 Jinguo Ye×1 Chao Li×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.

Bibliometrics come from a single OpenAlex record with no ORCID; this profile is name-resolved from public bibliographic data and 'Wenhao Wang' is a common name, so namesake contributions may be mixed in.
Citation impact could not be scored: only h-index 1 and i10-index 0 are available, with no per-paper citation counts.
Only two Evolve-reviewed papers underpin the qualitative assessment; the other 2026 works are counted in the totals but not reviewed.
countsByYear covers only 2026 (9 works), so career-length output trends cannot be established.