UCL machine learning researcher working on scalable Gaussian process surrogates and calibrated Bayesian optimisation.
2
papers here
1
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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 aggregate output metrics — total publication counts, venue records, or preprint volume over time — are available in this snapshot, so this axis is left unscored. It would measure the volume and consistency of research outputs across a career.
Impact—
Citation counts, h-index, and per-paper influence measures are absent from the input, so impact cannot be scored. This axis would measure how far the two catalogued 2026 preprints have propagated through the Bayesian optimisation and Gaussian process literature.
Focus100/100
The raw field distribution contains a single field — machine learning (Bayesian optimisation, scalable Gaussian processes) — and both 2026 catalogue papers fall inside it, producing the maximal focus score of 100. The agenda advances along one axis: product-of-experts Gaussian process surrogates, first deployed for calibrated Bayesian optimisation, then for information-based calibration of the uncertainty they quantify.
Collaboration28/100
The snapshot records 3 distinct coauthors, giving a low collaboration score of 28. Paolo Barucca, Wei Pan, and Jun Wang recur as frequent coauthors, and Ong is first author on both catalogued papers, so the output appears to come from one small, stable team rather than a broad, multi-group network.
Activity—
Career-span publication years are too thin to score activity: the snapshot holds only 2026-dated entries. This axis would measure output frequency and its trend over recent years.
Reproducibility55/100
Reproducibility averages 55 over 1 assessed paper. Evolve rates the Bayesian-optimisation paper's readiness as medium: the reported ~39% overhead cut against a global GP comes from synthetic-only tests, and regret gains are flagged as tiny and statistically unsupported, so replication on real tasks remains open.
Research summary
Yean Hoon Ong is a machine learning researcher at University College London whose catalogued work centres on making Gaussian process surrogates both scalable and trustworthy for Bayesian optimisation. The Evolve catalogue lists two 2026 arXiv papers, on both of which Ong is first author. The first, BO-pro-c, embeds a calibrated product-of-experts GP inside the BO loop; Evolve's review reports a real ~39% overhead cut versus a global GP in synthetic-only tests, while noting that regret improvements were tiny and statistically unsupported. The second proposes information-based calibration of uncertainty quantification in product-of-experts Gaussian process models, extending the same agenda from efficiency to how uncertainty itself is measured. This single-field concentration yields a focus score of 100: every catalogued output sits in machine learning (Bayesian optimisation, scalable Gaussian processes). Collaboration is compact — 3 distinct coauthors (collaboration score 28) — with Paolo Barucca, Wei Pan, and Jun Wang recurring. Reproducibility scores 55, based on one assessed paper with medium readiness. Output, impact, and activity axes lack the underlying bibliometric data and are left unscored, making this an early-stage profile that should be read as a snapshot of an emerging research line rather than an established record.
Collaboration pattern
Collaboration is compact: 3 distinct coauthors and a collaboration score of 28 point to a small working group. Paolo Barucca, Wei Pan, and Jun Wang are the recurring coauthors in the snapshot, and Ong takes first author position on both catalogued 2026 papers (neither as corresponding author). The pattern suggests a stable single-team pipeline around scalable, calibrated Gaussian processes rather than distributed multi-group work; broader or international collaborations are not visible in the available data.
Research keywords
Highlights
Calibrated product-of-experts GP surrogates cut BO overhead by ~39%
Information-based calibration of uncertainty quantification in PoE GP models
A second first-author 2026 arXiv paper proposes information-based calibration of uncertainty quantification in product-of-experts Gaussian process models; it awaits Evolve review, keywords, and a reproducibility assessment.
A single-problem, first-author research line
Timeline
- 2026First-author arXiv paper on Bayesian optimisation with calibrated product-of-experts Gaussian process models, reviewed by Evolve with medium reproducibility readiness.
- 2026First-author arXiv paper proposing information-based calibration of uncertainty quantification in product-of-experts Gaussian process models; not yet reviewed in the catalogue.
- 2026Both catalogued outputs fall in this single year, marking an early-stage and highly concentrated publication 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.
- Dimension scores are computed from a public bibliometric snapshot; citation counts, h-index, and venue histories were not provided.
- Coverage is limited to two 2026 arXiv preprints in the Evolve catalogue; peer-reviewed publications, funding, and career history fall outside this data scope.
- The second paper has no Evolve review, keywords, or reproducibility assessment yet, so quality signals rest on one reviewed paper.
- The profile is name-resolved from public bibliographic data (ORCID 0000-0002-8452-0999); country-level affiliation is not recorded in the input.