Jilin Hu

East China Normal University

Unclaimed · AI-structured profileOpenAlex

Time series forecasting researcher at East China Normal University: wind power benchmarks, frozen-backbone adaptation.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-13) 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 axis measures cumulative scholarly output — total publications, venue record, annual productivity — which the current snapshot lacks; only two Evolve-reviewed 2026 arXiv papers are on file.

Impact—

Not yet scored: citation counts, h-index and downstream influence are absent from the current data. This axis would measure how widely WPBench and MUSE are cited and reused by the forecasting community.

Focus0/100

Scores 0/100, computed from an even 50/50 split in the field distribution between machine learning for time series forecasting and its renewable-energy (wind power) specialization. In practice the two reviewed 2026 papers — a wind-power benchmark and a general multivariate forecasting method — sit in adjacent subfields of one theme, so the low score reflects breadth within a single line of work rather than scattered interests.

Collaboration53/100

Scores 53/100, built from 13 distinct coauthors across the cataloged record. Both reviewed papers carry non-first authorships — middle on WPBench, last on MUSE — without corresponding-author roles, and recurring names such as Xinying Cai, Yuhan Zhu, Junkai Lu and Xiangfei Qiu point to a stable in-house team at East China Normal University rather than a broad external network.

Activity—

Not yet scored: with only two 2026 arXiv entries on file, publication cadence and multi-year productivity trends cannot be computed. This axis would measure how consistently research output appears over time and across venues.

Reproducibility78/100

Scores 78/100, the mean repro-readiness across the two reviewed papers: MUSE is rated high and WPBench medium. The methodological paper reports results on 10 datasets, while the wind benchmark aggregates 26 datasets and 19 models from five families with shape and structure diagnostics; Evolve notes the benchmark lacks significance tests and visible code, keeping overall openness solid but incomplete.

Research summary

Jilin Hu is a researcher at East China Normal University working on machine learning for time series forecasting. The Evolve catalog splits this early record evenly between two adjacent subfields — general multivariate time series forecasting and its renewable-energy specialization in wind power (50/50 field distribution). Two 2026 arXiv papers have been reviewed. WPBench, where Hu holds a middle authorship, unifies 26 wind power datasets and 19 models from five families with shape-aware (DTW) and structure diagnostics; Evolve finds it well-scoped but flags missing significance tests and no visible code. MUSE, with Hu as last author, adapts a frozen masked-autoencoder vision backbone to multivariate forecasting through VCR/TPR adapters, reporting state-of-the-art results on 10 datasets; its reproducibility readiness is rated high, though reported gains are often thin and lack significance testing. Quantitatively, the profile records 13 distinct coauthors (collaboration 53/100), a mean repro-readiness of 78/100 across two papers, and a focus score of 0 driven by the even split between closely related field labels. With no citation, output-volume or activity data available, this is an early-stage profile resolved from public bibliographic data; 'Jilin Hu' is a common name, so identity confidence rests on the institutional affiliation recorded in the snapshot.

Collaboration pattern

Hu's two reviewed papers involve 13 distinct coauthors (collaboration score 53/100), with middle authorship on WPBench and last authorship on MUSE and no corresponding-author roles. Recurring partners such as Xinying Cai, Yuhan Zhu, Junkai Lu, Xiangfei Qiu and Yingshan Li point to a stable in-house team at East China Normal University, typical of large multi-author benchmark and method papers.

Research keywords

time series forecastingmultivariate forecastingwind power forecastingforecasting benchmarkstime series foundation modelsspatio-temporal forecastingshape-aware evaluationdynamic time warping (DTW)frozen vision backbonemasked autoencoder (MAE)parameter-efficient adaptationcross-variable dependency

Highlights

WPBench unifies 26 wind datasets and 19 models under one benchmark
Evolve's verdict: WPBench brings 26 wind power datasets and 19 models from five families into a single evaluation with shape-aware (DTW) and structure diagnostics — well-scoped, but it publishes no significance tests and has no visible code (repro readiness: medium).
MUSE adapts a frozen MAE vision backbone to multivariate forecasting
As last author, Hu contributed to MUSE, which introduces VCR/TPR adapters for parameter-efficient, dependency-aware adaptation of a frozen masked autoencoder, reporting state-of-the-art results on 10 datasets; Evolve rates repro readiness high but finds gains often thin and untested for significance.
A paired strategy: one benchmark, one method in the same year
Both 2026 papers form a coherent pair — a wind-domain evaluation infrastructure (WPBench) plus a general forecasting architecture (MUSE) — which explains the 50/50 field distribution and the 0/100 focus score recorded for this early profile.

Timeline

2026WPBench released on arXiv: a comprehensive wind power forecasting benchmark spanning 26 datasets and 19 models from five families, with shape-aware (DTW) diagnostics; Hu contributes as a middle author.
2026MUSE released on arXiv: dependency-aware VCR/TPR adapters adapt a frozen MAE vision backbone to multivariate forecasting with state-of-the-art results on 10 datasets; Hu is the last author.
2026Both cataloged papers land in the same year, establishing an early-stage record in time series forecasting with a mean repro-readiness of 78/100.

Papers in the Evolve catalog

Frequent co-authors

Xinying Cai×2 Yuhan Zhu×2 Junkai Lu×1 Xiaoyun Yu×1 Xiangfei Qiu×1 Yuhan Zhu×1 Yingshan Li×1 Li Ma×1 Xiangfei Qiu Linsen Li×1 Kai Zhang×1 Yao Fu×1 Weihao Jiang×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.

Bibliometric scores come from a small snapshot: only two arXiv papers reviewed by Evolve feed all dimension scores.
No citation, publication-volume or activity data is available; the output, impact and activity axes are unscored.
ORCID and country are not recorded, and 'Jilin Hu' is a common name; identity is resolved from public bibliographic data and should be treated as provisional.
Both papers are 2026 arXiv preprints; peer-reviewed venue acceptance is not reflected in the current data.