Ran He (CAS, h-index 63): face recognition and GANs veteran now on vision-language calibration and LLM post-training.
3
papers here
3
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
63
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—
Publication-output metrics are not available in this snapshot, so this axis is unscored. It would normally measure productivity — total works, first/corresponding-author share, and venue distribution — derived from the full indexed record rather than the three reviewed papers.
Impact—
Citation-based impact metrics are absent from this snapshot, so no score is given. This axis would measure influence through aggregate citations, citing works, and per-paper citation trajectories across the researcher's indexed record.
Focus0/100
The focus score is 0, reflecting wide dispersion rather than narrow specialization. The field distribution splits evenly across three areas — multi-domain RL for LLM post-training, computer vision/machine learning, and vision-language model calibration — while indexed topics span face recognition and analysis (147 works), generative adversarial networks and image synthesis (125), and domain adaptation and few-shot learning (64).
Collaboration53/100
The collaboration score is 53, a mid-range value. The current snapshot lists 13 distinct coauthors, including recurring partners such as Jian Liang and Tieniu Tan. Across the three Evolve-reviewed 2026 papers, Ran He appears as last author twice and middle author once, consistent with a senior or supervisory authorship role on recent work.
Activity100/100
Activity reaches the maximum score of 100. Indexed works climb from 1 in 2005 to a 2019 peak of 50, with 44 in 2021 and a steady 25–28 per year from 2022 onward; the last three years total 77 works, and 22 are already indexed for 2026 — a sustained, high-volume record spanning two decades.
Reproducibility85/100
Reproducibility averages 85 across the three Evolve-reviewed papers. Two carry high readiness — the temperature-scaling calibration study and the selective test-time adaptation work — while the multi-domain RL controller (OSOL) is rated medium, with careful theory and diagnostics but modest gains and no reported variance.
Research summary
Ran He is a researcher at the Chinese Academy of Sciences whose OpenAlex-resolved record spans two decades of computer vision and machine learning research. The record reports an h-index of 63 and an i10-index of 184, with yearly output rising from one indexed work in 2005 to a peak of 50 in 2019 and staying high since: 77 works over the last three years and 22 already indexed for 2026, yielding a maximum activity score of 100. Research breadth is considerable rather than narrow — the focus score is 0 — with indexed topics led by face recognition and analysis (147 works), generative adversarial networks and image synthesis (125), face and expression recognition (99), advanced neural network applications (88), and domain adaptation and few-shot learning (64). The Evolve-reviewed catalog captures a current turn toward vision-language models and LLM post-training: temperature scaling for calibrating test-time prompt tuning, selective test-time adaptation for CLIP-style models, and the OSOL cross-step controller for multi-domain GRPO. Reproducibility readiness averages 85 across these three reviewed papers (two high, one medium), while collaboration scores a moderate 53 on 13 distinct coauthors in the current snapshot, including recurring partners such as Jian Liang and Tieniu Tan.
Collaboration pattern
Collaboration is mid-range (score 53): 13 distinct coauthors appear in the current snapshot. Frequent coauthors include Jian Liang, Tieniu Tan, Guojun Yin, Wei Lin, Dapeng Hu, Yuwei Liang, Zihan Lin, Xiaohan Wang, Jie Cao, Jiajun Chai, and Siru Jiang. In the three reviewed 2026 papers, Ran He appears as last author twice and middle author once, consistent with a senior or supervisory authorship role.
Research keywords
Highlights
Calibrating test-time prompt tuning with zero-shot-anchored temperature scaling
Per-sample selective test-time adaptation for vision-language models
OSOL: cross-step control against higher-order interference in multi-domain RL
Timeline
- 2005Earliest indexed works appear in the OpenAlex record with 1 publication, marking the start of a two-decade output history.
- 2019Output peaks at 50 indexed works, the most productive single year in the record.
- 2021A second high point with 44 indexed works, close behind the 2019 peak.
- 2025Recent cadence holds steady at 28 indexed works, within the 25–28 annual range maintained since 2022.
- 2026Three arXiv papers enter the Evolve catalog: zero-shot-anchored temperature scaling for TPT calibration, selective test-time adaptation, and the OSOL multi-domain RL controller.
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.
- Bibliometric indicators (h-index 63, i10-index 184, yearly counts, topics) come from OpenAlex-sourced data; this profile is name-resolved from public bibliographic records.
- Output and impact dimensions are unscored because the underlying metrics were unavailable in this snapshot.
- Only three papers have been reviewed by Evolve; verdicts cover those 2026 arXiv preprints, not the full publication record.
- “Ran He” is a common Chinese name; identity is anchored to ORCID 0000-0002-3807-991X and the OpenAlex record.