Huiyuan Liu
Powerchina Huadong Engineering Corporation (China) · China
Audio-language-model researcher at Powerchina Huadong building refreshable benchmarks for telecom fraud detection
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-19) 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 cannot be scored yet: the snapshot contains no verified full publication record under this identity, only two 2026 arXiv works. This axis would measure total peer-reviewed output, venue quality, and year-over-year production volume once record coverage improves.
Impact—
Impact cannot be scored yet: citation aggregates are absent (h-index and i10-index both reported as 0) and the two 2026 preprints are too recent for citations to accumulate. This axis would measure citations, citing works, and uptake of the TeleAntiFrad 2.0 benchmark.
Focus0/100
Focus scores 0 in this snapshot. The two 2026 works split evenly (0.5/0.5) across two adjacent field labels — speech processing and NLP, and speech and audio machine learning — both aimed at fraud detection, while OpenAlex topic tags ('Impact of Technology on Adolescents', 'Innovative Human-Technology Interaction', 'Digital Mental Health Interventions', 2 counts each) do not match the reviewed papers. With recent3yWorks at just 2, thematic concentration cannot yet be established.
Collaboration58/100
Collaboration scores 58. Across the two 2026 papers, OpenAlex counts 17 distinct coauthors, and 12 recur as frequent collaborators, including Shun Zhang, Qifan Wang, and Yifan Wang. Liu's roles differ by project — first author on TeleAntiFrad 2.0, middle author on FRAUDSkill — and neither paper lists Liu as corresponding author, indicating contribution within a larger team.
Activity32/100
Activity scores 32. The countsByYear record contains only 2026 (2 works) and recent3yWorks is 2, with no earlier years represented in the snapshot; h-index and i10-index are both 0. The visible pattern is a concentrated recent burst — two arXiv preprints in a single year — rather than sustained multi-year production.
Reproducibility78/100
Reproducibility scores 78, a mean computed across 2 scored papers. TeleAntiFrad 2.0 is rated high readiness, reflecting its refreshable benchmark design, near-domain negatives, and collapse-aware analysis, though its data is synthetic-only; FRAUDSkill is rated medium, with Evolve's ablation review attributing most of the +31.96 Macro-F1 gain to structured post-processing rather than skill search.
Research summary
Huiyuan Liu is a researcher at Powerchina Huadong Engineering Corporation in China whose recent published work centers on audio-language models for telecom fraud detection. The OpenAlex-based snapshot records two works in the last three years, both posted to arXiv in 2026, with field distribution split evenly (0.5/0.5) between speech processing and NLP and speech and audio machine learning, both oriented to fraud detection. Liu is first author of TeleAntiFrad 2.0, a refreshable, profile-grounded, audio-based benchmark that introduces near-domain negatives and a collapse-aware analysis of shortcut learning; Evolve rates its reproducibility readiness as high while flagging synthetic-only data and thin human validation. Liu appears as a middle author on FRAUDSkill, a frozen-weight skill-optimization method for audio anti-fraud detection whose reported +31.96 Macro-F1 gain Evolve's ablation review attributes largely to structured post-processing rather than skill search. Across the two papers the platform counts 17 distinct coauthors, a mean reproducibility score of 78 (2 papers scored), an activity score of 32, and h-index and i10-index of 0 — consistent with a newly visible, early-stage record. Output and impact axes are not yet scorable from the available data.
Collaboration pattern
Liu's visible work is team-produced: 17 distinct coauthors across two 2026 papers, with 12 recurring frequent collaborators including Shun Zhang, Qifan Wang, Yifan Wang, Yijin Zhou, and Yuxi Zhao. Authorship placement shifts with the project — first author on the TeleAntiFrad 2.0 benchmark, middle author on FRAUDSkill — and neither position is marked corresponding, indicating a contributor role within a larger engineering group rather than an independent lab lead.
Research keywords
Highlights
TeleAntiFrad 2.0: a refreshable audio benchmark for telecom fraud detection
FRAUDSkill: frozen-weight skill optimization with a contested +31.96 Macro-F1 gain
A benchmark-plus-method pairing on the same anti-fraud problem
Both 2026 works target audio-based telecom fraud detection — one supplying a refreshable benchmark (first author), the other a frozen-model optimization method (middle author) — with 17 distinct coauthors across the pair and a mean reproducibility score of 78 over 2 scored papers.
Timeline
- 2026Posts TeleAntiFrad 2.0 to arXiv as first author — a refreshable, profile-grounded, audio-based benchmark for telecom fraud detection featuring near-domain negatives and collapse-aware analysis.
- 2026Co-authors FRAUDSkill on arXiv, a structured frozen-weight skill-optimization method for audio anti-fraud detection, as a middle author.
- 2026OpenAlex records 2 works in 2026 with 17 distinct coauthors, a mean reproducibility score of 78 across 2 scored papers, and an activity score of 32.
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 figures come from OpenAlex only; no ORCID is provided, so the record is name-resolved and may conflate researchers sharing the name Huiyuan Liu.
- Output and impact dimensions are unscored: beyond h-index 0 and i10-index 0, no citation counts are available, and both works are 2026 preprints too recent for citations.
- OpenAlex topic labels (Impact of Technology on Adolescents; Innovative Human-Technology Interaction; Digital Mental Health Interventions) do not match the reviewed fraud-detection papers, indicating topic-assignment noise.
- The evidence base is limited to two 2026 arXiv preprints reviewed by Evolve; peer-reviewed publications, if any, are outside this snapshot.