Chen Hou

Minjiang University

Unclaimed · AI-structured profileOpenAlex

Chen Hou (Minjiang University): early-career researcher spanning LLM-agent privacy leakage and time-series forecasting

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-14) 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 scored: this snapshot provides no publication counts, venue list, or yearly output totals. The axis would measure the volume, consistency, and venues of Chen Hou's research output over time.

Impact—

Insufficient data: no citation counts, h-index, or downstream-adoption signals were provided. The axis would measure how often and how widely Chen Hou's work is cited and built upon.

Focus0/100

The radar focus score is 0, derived from an even 50/50 field distribution across machine learning (time-series forecasting) and AI security & privacy (LLM agents). With only two 2026 arXiv preprints — a plug-in covariate gate for Transformer forecasters and the CIPL privacy-leakage framework for LLM agents — the profile reads as breadth-oriented and exploratory rather than concentrated in a single research line.

Collaboration44/100

Collaboration scores 44, built on 8 distinct coauthors across the two reviewed preprints: Tao Huang, Hongkai Zhuang, Guosen Wu, Guolong Zheng, Jiayang Meng, Xu Yang, Xuechao Yang, and Feng Xia. Chen Hou sits in a middle author position on the CIPL privacy paper and in the last position on the forecasting-gate paper, with neither marked corresponding — a moderate, team-embedded pattern rather than clear independent leadership.

Activity—

No data yet: beyond the two 2026 preprints there is no publication frequency, submission cadence, or multi-year activity record. The axis would measure research tempo and sustained engagement over time.

Reproducibility55/100

Reproducibility is the strongest scored axis at 55, the mean across 2 reviewed papers, both rated medium readiness. Evolve's review of CIPL finds results coherent across 5 LLM providers but without statistical testing; the forecasting-gate paper is honestly framed with mostly parity results yet likewise lacks significance tests — transparent, consistent work that is not yet statistically hardened.

Research summary

Chen Hou is a researcher at Minjiang University whose early, name-resolved bibliographic profile currently rests on two 2026 arXiv preprints in distinct subfields. The first, CIPL, formalizes exposure-versus-recovery privacy leakage in black-box LLM agents, with indexed keywords spanning RAG privacy, tool-use security, and memory extraction, and results reported as coherent across 5 providers; Evolve's review finds them coherent but unaccompanied by statistical testing. The second proposes a lightweight plug-in gate — a 2-layer MLP reweighting covariates before the encoder — for 3 Transformer-based long-term time-series forecasters, reporting mostly parity results with honest framing and, again, no significance tests. The reviewed output splits evenly (50/50) between machine learning (time-series forecasting) and AI security & privacy (LLM agents), yielding a radar focus score of 0 and marking the profile as exploratory rather than specialized. Collaboration is moderate (score 44), with 8 distinct coauthors including Tao Huang, Hongkai Zhuang, Guosen Wu, Guolong Zheng, Jiayang Meng, Xu Yang, Xuechao Yang, and Feng Xia; authorship positions are middle on CIPL and last on the forecasting paper, neither corresponding. Reproducibility is the strongest scored dimension (55, the mean over 2 papers, both rated medium readiness). Output, impact, and activity axes remain unscored because citation and volume data were not provided, so this should be read as an early-stage profile.

Collaboration pattern

The reviewed preprints involve 8 distinct coauthors — Tao Huang, Hongkai Zhuang, Guosen Wu, Guolong Zheng, Jiayang Meng, Xu Yang, Xuechao Yang, and Feng Xia — yielding a moderate collaboration score of 44. Chen Hou holds a middle author position on the CIPL privacy paper and the last position on the forecasting-gate paper, with neither marked corresponding, suggesting a team-embedded contribution pattern rather than independent leadership; longer-term co-authorship persistence cannot be assessed from two papers.

Research keywords

LLM agentsprivacy leakageblack-box evaluationRAG privacytool-use securitymemory extractionlong-term time-series forecastingTransformercovariate admission gateplug-in modulefeature selection

Highlights

CIPL: formalizing exposure-vs-recovery privacy leakage in black-box LLM agents
This 2026 arXiv preprint formalizes how private content exposed to black-box LLM agents can be recovered, with indexed keywords spanning RAG privacy, tool-use security, and memory extraction. Evolve's verdict: results are coherent across 5 providers but lack statistical testing; reproducibility readiness is rated medium.
A lightweight covariate-admission gate for Transformer forecasters
A 2-layer MLP plug-in module reweights covariates before the encoder in 3 Transformer-based long-term time-series forecasters, with usage regularization for feature selection. Evolve finds results mostly at parity with honest framing, but no significance tests; reproducibility readiness is medium.

Timeline

2026Releases CIPL, a channel-aware framework for recoverable privacy leakage in black-box LLM agents, as an arXiv preprint in a middle author position (not corresponding).
2026Releases a lightweight plug-in covariate gate for Transformer-based long-term time-series forecasters as an arXiv preprint in the last author position (not corresponding).
2026Both preprints enter the Evolve catalog with medium reproducibility readiness, splitting the reviewed profile evenly (50/50) between time-series forecasting and LLM-agent privacy.

Papers in the Evolve catalog

Frequent co-authors

Tao Huang×2 Hongkai Zhuang×1 Guosen Wu×1 Guolong Zheng×1 Jiayang Meng×1 Xu Yang×1 Xuechao Yang×1 Feng Xia×1

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.

Profile is name-resolved from public bibliographic data; 'Chen Hou' is a common name and no ORCID was provided, so identity matching carries some uncertainty.
Bibliometric dimensions come from a small snapshot of only 2 reviewed 2026 arXiv preprints; no citation counts, h-index, or publication-volume data were provided, so output, impact, and activity remain unscored.
Both papers are arXiv preprints; peer-review status and venue acceptance are unknown.
Country and the Chinese-character form of the name are unverified; the institutional affiliation is taken as given in the source data.