Rui Xie

University of Central Florida

Unclaimed · AI-structured profileOpenAlex

Rui Xie (UCF): verifiable environments for computer-use agents, statistical-control theories of LLM problem solving

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-14) 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 publication-volume or output data is available in the current snapshot, so this axis is unscored. It would measure the volume, cadence, and venue spread of Rui Xie's papers and preprints over time.

Impact—

Citation counts, h-index, and downstream-adoption signals were not provided, so impact cannot be scored. This axis would measure how widely the two 2026 preprints and any other outputs are cited, reused, or built upon.

Focus0/100

The focus score is 0, reflecting an even 0.5/0.5 split across two fields: machine learning / LLM agents and machine learning / statistics (stat.ML). The two 2026 papers embody this duality — one builds verifiable benchmark environments for computer-use agents, the other reframes LLM complex problem solving in statistical-control terms. Topical concentration is therefore minimal in this snapshot: the portfolio is deliberately two-sided rather than narrowly focused.

Collaboration85/100

The collaboration score is 85, drawn from 69 distinct coauthors in the record. Twelve researchers recur as frequent collaborators, including Shuai Bai, Jiayong Deng, and Sicheng Fan. Both reviewed 2026 papers are large-team arXiv efforts in which Rui Xie holds a middle authorship position — not first or corresponding author — a pattern consistent with high-throughput, team-based machine learning research.

Activity—

Beyond the two 2026 catalog entries, no publication-cadence data was provided, so activity is unscored. This axis would measure annual output rhythm, recent momentum, and sustained engagement across venues over multiple years.

Reproducibility60/100

Reproducibility averages 60 across 2 scored papers, with a sharp internal contrast: RecreationWorld is rated high readiness thanks to its verifiable five-platform environment design, while the statistical-control survey is rated low because its diagnostic hypotheses remain untested. Reproducible practice currently looks stronger on artifact-building work than on theory-only writing.

Research summary

Rui Xie is a machine learning researcher at the University of Central Florida whose recorded work sits at the intersection of LLM agents and statistical methodology. The available snapshot resolves evenly across two fields — machine learning / LLM agents and machine learning / statistics (stat.ML), each at 0.5 — and this duality is visible in the two 2026 arXiv papers reviewed by Evolve. The first, RecreationWorld, contributes scalable and verifiable environments for hybrid computer-use agents: a five-platform recreation benchmark paired with an SFT pipeline for GUI and coding agents; Evolve's review calls the evaluation design rigorous while noting that transfer gains lack statistical controls, and rates its reproducibility readiness high. The second reframes complex problem solving in large language models as sequence estimation and control, drawing on statistical decision theory, Bayes filters, and uncertainty quantification; it is judged well-organized and candid, though its diagnostic hypotheses remain untested and its reproducibility readiness is low. Collaboration is the strongest scored dimension at 85, with 69 distinct coauthors and 12 frequent collaborators, and Rui Xie appears in middle authorship positions on both papers rather than as first or corresponding author. Output, impact, and activity lack sufficient underlying data and are left unscored.

Collaboration pattern

The record shows 69 distinct coauthors and a tight recurring core of 12 frequent collaborators, including Shuai Bai, Jiayong Deng, and Sicheng Fan — a hub-and-team pattern typical of large benchmark-building projects. On both reviewed 2026 papers Rui Xie sits in a middle authorship slot and is not the corresponding author, suggesting contribution through specialized subsystems rather than sole leadership. The high collaboration score of 85 reflects breadth of ties rather than first-author independence.

Research keywords

computer-use agentsGUI agentshybrid coding agentsbenchmarkverifiable rewardsrecreation taskslarge language modelscomplex problem solvingprocess controlstatistical decision theoryBayes filteruncertainty quantification

Highlights

RecreationWorld: verifiable five-platform environments for hybrid computer-use agents
The 2026 preprint contributes a recreation benchmark spanning five platforms together with an SFT pipeline for hybrid GUI/coding agents. Evolve rates the evaluation design rigorous and the reproducibility readiness high, while flagging that reported transfer gains lack statistical controls.
A statistical-control lens on complex problem solving in LLMs
The second 2026 survey recasts complex problem solving in large language models as sequence estimation and control, organized around statistical decision theory, Bayes filters, and uncertainty quantification. Evolve finds it well-structured and candid, but its diagnostic hypotheses remain untested, and reproducibility readiness is rated low.
A methodological signature pairing verifiable artifacts with control theory
Across both 2026 papers the same instinct appears twice: build verifiable evaluation surfaces, then describe agent behavior in estimation-and-control terms. Both Evolve verdicts also flag the same open gap — absent statistical controls and untested diagnostic hypotheses — marking a natural next step for this line of work.

Timeline

2026RecreationWorld appears on arXiv, introducing a five-platform recreation benchmark and an SFT pipeline for hybrid GUI/coding agents.
2026A statistical-control survey of complex problem solving in LLMs appears on arXiv, framing the task as sequence estimation and control.
2026The collaboration record reaches 69 distinct coauthors with 12 frequent collaborators, while research remains split evenly between LLM-agent and statistics work.

Papers in the Evolve catalog

Frequent co-authors

Shuai Bai×1 Jiayong Deng×1 Sicheng Fan×1 Yikun Fu×1 Chang Gao×1 Xuhao Hu×1 Mianqiu Huang×1 Yizhen Jiang×1 Yuheng Jing×1 Dehui Kong×1 Keliang Li×1 Ning 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.

'Rui Xie' is a common name; this profile is name-resolved from public bibliographic records tied to University of Central Florida, and same-name authors may have been merged or missed.
Quantitative signals come from public bibliometric data as processed by Evolve; no citation counts, h-index, or venue-level output totals were provided.
Qualitative verdicts cover only the two 2026 arXiv preprints in the Evolve catalog; earlier or peer-reviewed work is not represented.
No ORCID is linked and country-level affiliation data is absent, so disambiguation rests on institution metadata alone; output, impact, and activity axes are unscored for lack of data.