Machine learning researcher at Guangzhou University working on federated long-tail learning and machine unlearning.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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 scoreable. OpenAlex returns no verifiable cumulative publication record for this identity beyond the two 2026 arXiv papers Evolve has catalogued (h-index 0, i10-index 0), so overall research productivity cannot be measured yet.
Impact—
Not yet scoreable. Reported h-index and i10-index are both 0 and no citation counts are available, so this axis — which would measure citation uptake, influence on follow-up work, and downstream adoption — cannot be estimated.
Focus0/100
Focus scores 0, reflecting a deliberately split agenda: the field distribution is an even 0.5 machine learning (federated learning) and 0.5 machine learning (trustworthy ML / medical imaging), and OpenAlex topics scatter across five clusters — privacy-preserving technologies (2), domain adaptation and few-shot learning, AI in healthcare and education, explainable AI, and data classification (1 each). Both catalog papers do share class imbalance as a common thread.
Collaboration22/100
Collaboration scores 22 on a thin network: OpenAlex records only 2 distinct co-authors in the recent window, with Minghui Huang and Changkun Lin recurring across the catalogued work. Both 2026 arXiv papers place Junxiao Wang in the last author position without corresponding-author status, suggesting a contributing role within a small team rather than an independently led group.
Activity32/100
Activity scores 32, consistent with limited recent volume: OpenAlex counts only 2 works in the last three years, and the counts-by-year series contains a single entry (2026: 2 papers). Both appear as arXiv preprints in the Evolve catalog in 2026, so current output is real but early-stage and thin.
Reproducibility55/100
Reproducibility scores 55, the mean over the 2 papers Evolve has reviewed, both rated 'medium' readiness. FedLTLib fixes Dirichlet α = 0.5, reports no error bars, and shows claim-table gaps; GRIN+'s BUSI results (RetDev 94.2, R-Acc 58.3%) contradict its claimed PEU balance. The released benchmark and method give a workable but unpolished basis for replication.
Research summary
Junxiao Wang is a machine learning researcher at Guangzhou University (China) whose recent work concentrates on two branches of trustworthy machine learning: federated learning and machine unlearning for medical imaging. The OpenAlex-derived snapshot splits the profile evenly between these areas — machine learning (federated learning) and machine learning (trustworthy ML / medical imaging) each account for 0.5 of the field distribution — with topic clustering around privacy-preserving technologies in data (2 works), domain adaptation and few-shot learning, AI in healthcare and education, explainable AI, and machine learning and data classification. The Evolve catalog holds two 2026 arXiv papers, both authored in the last position: FedLTLib, a benchmark unifying 13 federated long-tail learning algorithms under a decoupled imbalance/skew taxonomy, and GRIN+, a gradient-ratio one-shot unlearning method for imbalanced medical data. Evolve rates both 'medium' on reproducibility readiness, giving a reproducibility mean of 55 across 2 reviewed papers. Output and impact are not yet scoreable (h-index 0, i10-index 0), and activity indicators are thin: 2 works in the last three years and 2 distinct co-authors, Minghui Huang and Changkun Lin.
Collaboration pattern
Collaboration is small-scale: OpenAlex records 2 distinct co-authors, with Minghui Huang and Changkun Lin recurring across the catalogued work. In both 2026 arXiv papers Junxiao Wang sits in the last author position and is not the corresponding author, consistent with a contributing role inside a compact team rather than leading an independent group. The collaboration score of 22 reflects this narrow, stable core rather than extended co-author networks.
Research keywords
Highlights
FedLTLib: a unified benchmark for federated long-tail learning
GRIN+: one-shot machine unlearning for imbalanced medical data
Timeline
- 2026FedLTLib benchmark preprint released on arXiv, unifying 13 federated long-tail learning algorithms under a decoupled imbalance/skew taxonomy.
- 2026GRIN+ preprint released on arXiv, proposing gradient-ratio one-shot machine unlearning for imbalanced medical data.
- 2026Both papers enter the Evolve catalog in last-author position with 'medium' reproducibility readiness (reproMean 55 across 2 papers).
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.
- Bibliometrics are derived from OpenAlex only; no ORCID is linked, so this profile is name-resolved from public bibliographic data and 'Junxiao Wang' may be ambiguous.
- Output and impact are unscored: reported h-index and i10-index are both 0, and no citation or venue history beyond the two catalogued 2026 arXiv papers is available.
- The reproducibility score (mean 55) covers only the 2 papers Evolve has reviewed.
- Activity indicators are thin (2 works in the last three years), so career-stage or trajectory conclusions should not be drawn.