Amazon-based researcher across multimodal vision-language models, image quality assessment, and industrial graph ML
3
papers here
3
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
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—
No publication-count or output metrics are available for this profile yet. This axis would measure total scholarly output, venue distribution, and productivity over time; current evidence is limited to three 2026 arXiv preprints in the Evolve catalog.
Impact—
Citation-based impact data is not yet available, so this axis is not scored. It would measure citation counts, h-index-style influence, and uptake of the three reviewed papers across the image quality, multimodal, and graph communities.
Focus0/100
The focus score of 0 reflects an unusually even spread: the raw field distribution assigns exactly one third each to graph machine learning for industrial fraud detection, multimodal vision-language machine learning, and computer vision for image quality assessment. All three 2026 catalog papers sit in different fields, so no dominant research line emerges within the reviewed sample.
Collaboration62/100
Scoring 62, this axis rests on 21 distinct coauthors in the reviewed work, with a recurring core of twelve frequent collaborators such as Yice Luo, Yun Zhu, Yongchao Liu, Kai Zhang, and Simon Shaolei Du. All three catalog papers are middle-authored and non-corresponding, pointing to distributed team science inside an industry lab rather than solo-led projects.
Activity—
Annual publication cadence cannot be computed from the available data, so this axis is unscored. It would measure yearly output rhythm and research continuity; the only signal at hand is that all three reviewed papers are dated 2026.
Reproducibility43/100
The reproducibility mean of 43 is computed over three scored papers. Evolve rates the SIQA-U challenge report and the image tokenizer study as medium reproducibility readiness, and the GraphFAS industrial system as low, reflecting an untested baseline set and a deployment context that is hard to replicate.
Research summary
Xi Chen is a researcher at Amazon whose reviewed work spans three roughly equal areas: graph machine learning for industrial fraud detection, multimodal vision-language modeling, and computer vision focused on scientific image quality assessment. The Evolve catalog contains three 2026 arXiv papers, in all of which Chen is a middle author without corresponding-author status, consistent with contributor roles inside larger industry research teams. The most application-grounded item is GraphFAS, a CIKM'26 system paper describing a distributed, Boruta-based graph feature generation and selection pipeline reported as deployed on Alipay, with strong results but untested nearest baselines. On the multimodal side, a testbed study of image tokenizers in unified autoregressive models shows that task-specific validation loss — especially image-to-text — judges tokenizers better than reconstruction metrics, within a scope of up to 8B parameters, 256px resolution, and seven tokenizers. A challenge report on scientific image quality assessment claims first place in SIQA-U through multimodal retrieval-augmented generation built on ColPali and reciprocal rank fusion, though missing ablations and the validation-set indexing scheme raise data-leakage concerns. Bibliometric signals remain narrow: 21 distinct coauthors and a collaboration score of 62, an even one-third field split that yields a focus score of 0, and a reproducibility mean of 43 across three scored papers, while output, impact, and activity axes await sufficient data.
Collaboration pattern
Chen's reviewed output is team-based industry research: all three 2026 papers are middle-author, non-corresponding contributions. The collaboration score of 62 rests on 21 distinct coauthors, with a recurring core of twelve frequent collaborators — including Yice Luo, Yun Zhu, Yongchao Liu, Kai Zhang, and Simon Shaolei Du — spanning industrial system work (GraphFAS is attributed to Ant Group) and multimodal modeling studies.
Research keywords
Highlights
Claims first place in SIQA-U via multimodal retrieval-augmented generation
Shows validation loss beats reconstruction metrics for judging image tokenizers
GraphFAS: distributed graph feature automation reported deployed at Alipay
Timeline
- 2026Publishes a challenge report claiming first place in SIQA-U scientific image quality assessment using multimodal retrieval-augmented generation.
- 2026Co-authors a testbed study of image tokenizers as visual languages in unified multimodal models.
- 2026Contributes to GraphFAS, a CIKM'26 distributed graph feature system reported as deployed in industrial transaction networks.
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 dimensions are derived from OpenAlex-sourced data only; the output, impact, and activity axes lack sufficient data and are left unscored.
- Name resolution: Xi Chen is a common name and no ORCID is provided; this profile is resolved from public bibliographic records and may conflate distinct researchers.
- Evidence is limited to three 2026 arXiv preprints reviewed by Evolve; citation counts, venue acceptance records, and h-index data are outside scope.
- The reproducibility score (43) is Evolve's repro-readiness mean over three papers, not an independent replication audit.