ML researcher at Renmin University of China spanning tabular foundation models and vision-language GUI agents
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
2
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—
Output is unscored: this snapshot lacks a verified complete works list, venue distribution and per-paper output counts. The axis would measure total publications and their venues; only year-level counts (3 in 2025, 11 in 2026) are exposed here.
Impact—
Impact is unscored because citation data here are too sparse to interpret — only h-index 2 and i10-index 1 from a very recent output burst. The axis would measure citation counts, citing works and influence as papers accumulate citation history.
Focus0/100
The focus score is 0, driven by an even fieldDist split: 0.5 statistical machine learning (tabular data) versus 0.5 machine learning (multimodal vision-language GUI agents). Topic counts reinforce the breadth — multimodal applications (10), advanced neural networks (7), GANs and image synthesis (7), domain adaptation and few-shot learning (4), and NLP techniques (3) — so the portfolio spans two distinct method communities.
Collaboration65/100
The collaboration score of 65 rests on 25 distinct coauthors across recent work, with 12 recurring collaborators including Erbo Li, Ting Wei, Zizhou Huang and Zhangxuan Gu. Authorship roles are mixed: first author on the PACE tabular screening paper and a middle-position contributor on LLaDA-UI, indicating both independent-led projects and team-based systems work typical of larger vision-language groups.
Activity79/100
Activity scores 79 on the strength of 14 works in the recent three-year window and a sharp year-over-year rise in countsByYear, from 3 works in 2025 to 11 in 2026. Both catalogued papers are 2026 arXiv releases, consistent with an intense current output phase; the h-index of 2 confirms this volume is too new to have accumulated citation weight yet.
Reproducibility55/100
Reproducibility averages 55 across the 2 papers Evolve assessed, both rated medium readiness. PACE's controls were judged convincing, though its headline TALENT gains trace to nuisance-augmented tables; LLaDA-UI reports single-run evaluations without confidence intervals and a latency study with n=5. The mid-range score reflects solid experimental hygiene in one paper offset by statistical-reporting gaps in the other.
Research summary
Qi Qin is an early-career machine learning researcher at Renmin University of China whose recorded work splits evenly, by Evolve's field distribution, between statistical machine learning for tabular data and multimodal vision-language GUI agents. The tabular strand leads with PACE (2026, first author), a plug-and-play contextual-embedding method that reuses frozen column embeddings from pretrained tabular foundation models such as TabICL and TabPFN to upgrade classical feature screening via distance correlation; Evolve judges the controls convincing while cautioning that the headline TALENT gains stem from nuisance-augmented tables. The multimodal strand contributes to LLaDA-UI (2026, middle author), a block-wise diffusion language model for GUI agents reported to beat Qwen2.5-VL-7B on all six benchmarks and Qwen3-VL-8B on four of six, though with single-run evaluations and no confidence intervals. Bibliometric coverage is thin but rising fast: an h-index of 2 and i10-index of 1, 14 works in the recent three-year window, a jump from 3 works in 2025 to 11 in 2026, and 25 distinct coauthors. Topic signals point to multimodal applications, neural network applications, GAN-based image synthesis, few-shot learning and NLP. This is a fast-moving early-stage profile best read alongside its caveats.
Collaboration pattern
Qi Qin's collaboration footprint spans 25 distinct coauthors, with a recurring core of 12, including Erbo Li, Ting Wei, Zizhou Huang, Zixuan Qin, Wu Wang, Yifan Sun, Zhangxuan Gu, Haoxing Chen, Yi Xin, Kai Gan, Lin Liu and Long Cui. Roles alternate between leading (first author on PACE) and contributing within larger teams (middle author on LLaDA-UI), yielding a collaboration score of 65 — broad but not yet anchored by a single long-term consortium.
Research keywords
Highlights
First-author feature-screening method built on frozen tabular foundation models
Co-author on a block-wise diffusion vision-language GUI agent
Rapidly expanding dual-track output
Recorded output jumped from 3 works in 2025 to 11 in 2026, with 14 works in the recent three-year window and 25 distinct coauthors, while the field distribution splits evenly between tabular statistical ML and multimodal vision-language GUI agents — a broad, fast-scaling early portfolio.
Timeline
- 2025Earliest recorded output in the OpenAlex snapshot: 3 works attributed to 2025.
- 2026First author on PACE, applying frozen TabICL/TabPFN column embeddings to classical feature screening.
- 2026Middle-position co-author on LLaDA-UI, a block-wise diffusion vision-language GUI agent.
- 2026Recorded output jumped to 11 works in 2026 (from 3 in 2025), with 14 works across the recent three-year window.
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 come from an OpenAlex-based snapshot; no ORCID was available for identity verification, and 'Qi Qin' is a common Chinese name, so this profile is name-resolved from public bibliographic data.
- Citation-impact signals (h-index 2, i10-index 1) reflect brand-new output and are too early to interpret as influence.
- Only 2 papers have been assessed by Evolve; the other works behind the 14-work recent-3y count are not individually reviewed.
- Venue-level metadata and complete per-paper records are absent, which is why the output and impact axes remain unscored.