Yingcheng Wu
—
ML researcher working on JEPA-style world models and open-ended discovery intelligence for AI-driven science
2
papers here
2
reviewed by Evolve
0
first-author
0
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—
Insufficient data: this profile has no publication-volume metrics, so an output score cannot be computed. The axis would measure the number and cadence of indexed publications over time.
Impact—
Insufficient data: no citation counts or citation-based indicators are available yet. This axis would measure the influence of the work through citations and downstream adoption by other researchers.
Focus0/100
The focus score is 0 on the 0-100 scale, the lowest reading in this profile. The raw field distribution splits evenly — 0.5 artificial intelligence and 0.5 machine learning (self-supervised representation learning and world models) — across only two catalogued papers, one on JEPA-style world models and one a broad AI-for-science framework. With such a small corpus, this even split reads as breadth over concentration, and the score should be re-read as the catalog grows.
Collaboration55/100
The collaboration score is 55 out of 100, drawn from 15 distinct coauthors in the raw record. The frequent-coauthor list — including Ling Yang, Zhenfei Yin, Weiyang Liu, Mengyue Yang, and Wanli Ouyang — suggests work within a connected research cluster. On the two catalogued papers Wu is a middle-listed co-author on one and the last-listed (non-corresponding) author on the other, consistent with multi-author team science rather than solo output.
Activity—
Insufficient data: the snapshot lacks year-by-year publication records beyond the two 2026 preprints, so recent-activity level cannot be scored. This axis would measure publishing rate and recency of output across recent years.
Reproducibility38/100
The reproducibility score is a mean of 38, computed from exactly two assessed papers. Evolve rates JEPA-Anything at medium reproducibility readiness — gains span seven domains but vary in size, and wet-lab/clinical claims lack verifiable textual detail — while Discovery Foundation Models is rated low readiness, with nothing implemented, measured, or shown to transfer. Both ratings cap confidence in independent replication today.
Research summary
Yingcheng Wu is a machine learning researcher whose catalogued work to date sits at the intersection of self-supervised world models and AI-driven scientific discovery. The Evolve catalog lists two 2026 arXiv preprints. In 'JEPA-Anything: Learning Predictive Models across Different Worlds', Wu is a middle-listed co-author on work that adds orthogonal factor predictors to the JEPA (joint-embedding predictive architecture) framework; Evolve's review notes matched gains across seven domains, while flagging that gains vary in size and headline wet-lab and clinical claims lack verifiable textual detail. In 'Discovery Foundation Models: Toward Open-Ended Discovery Intelligence', Wu is the last-listed author (not corresponding) on a framework paper defining Discovery Foundation Models through a taxonomy and process protocol spanning research state, hypothesis revision, and experimental design — ambitious in scope, though nothing is yet implemented or measured. Bibliometric signals remain limited: 15 distinct coauthors, an even 50/50 split between artificial intelligence and machine learning (self-supervised representation learning and world models), a collaboration score of 55, a focus score of 0, and a reproducibility mean of 38 across two assessed papers. Output, impact, and activity scores are not yet computable from the available snapshot.
Collaboration pattern
Wu's recorded collaboration footprint spans 15 distinct coauthors, with a collaboration score of 55 out of 100. Recurring names on the frequent-coauthor list include Ling Yang, Zhenfei Yin, Taoyong Cui, Zhongyao Wang, Xinyue Xu, Weiyang Liu, Zhaochen Yu, Yuying Zhang, Qiang Gao, Mengyue Yang, and Wanli Ouyang. Both catalogued papers are multi-author efforts — Wu sits mid-list on the JEPA-Anything paper and last-listed (non-corresponding) on Discovery Foundation Models — pointing to team-based research within a stable cluster.
Research keywords
Highlights
Orthogonal factor predictors extend JEPA across seven domains
A framework paper defines Discovery Foundation Models
Timeline
- 2026Co-authors 'JEPA-Anything: Learning Predictive Models across Different Worlds' (arXiv preprint, middle-listed author)
- 2026Last-listed author on 'Discovery Foundation Models: Toward Open-Ended Discovery Intelligence' (arXiv preprint, non-corresponding)
- 2026Both catalogued preprints appear in 2026, pairing world-model architecture work with an AI-for-science framework
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.
- Quantitative scores are computed from a limited bibliometric snapshot; no citation counts, h-index, or full publication volume were available.
- No institutional affiliation, country, or ORCID was provided; the profile is name-resolved from public bibliographic data and may merge or split namesakes.
- Qualitative assessments rest on only two Evolve-reviewed papers (both 2026 arXiv preprints); the reproducibility mean of 38 is computed from just those two.
- Output, impact, and activity axes lack underlying data and are left unscored rather than estimated; career facts such as awards and funding are out of scope.