Early-stage NILM researcher: first-author arXiv work on training-only multi-appliance load disaggregation
1
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-20) 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—
Insufficient data: Evolve has reviewed only one arXiv preprint for this researcher, so total publication count, venue tiers, and yearly output volume cannot be scored. This axis would measure the number and cadence of peer-reviewed publications.
Impact—
Insufficient data: no citation counts, h-index, or downstream uptake are available in the input, and the single reviewed paper is a recent arXiv preprint. This axis would measure citations, adoption, and influence on subsequent NILM research.
Focus100/100
Score 100/100: every paper currently in the Evolve catalog — one of one — falls in machine learning for energy systems (NILM / smart grids), per the field distribution of reviewed works. The work targets multi-appliance energy disaggregation with domain generalization, data augmentation, and consistency regularization. With only one reviewed paper, this perfect concentration reflects a narrow visible record rather than a long-term specialization.
Collaboration14/100
Score 14/100 reflects a very small observed collaboration footprint: the reviewed paper lists one distinct co-author, Yanfang Fan, giving a low distinct-coauthor count in the raw bibliometrics. It is a first-author (non-corresponding) single-partner paper, so the low score may partly reflect an early-stage record with only one reviewed work rather than a stable solo working style.
Activity—
Insufficient data: only one paper, dated 2026, is present, so publication cadence, recent-year trends, and career-stage activity level cannot be computed. This axis would measure yearly output frequency and whether activity is accelerating, steady, or slowing.
Reproducibility55/100
Score 55/100 is the mean repro-readiness across one scored paper (reproPapers = 1). Evolve's verdict notes solid REDD gains but non-significant results on UK-DALE and REFIT, and no ablation studies or released code; readiness is rated medium. Reproducing the headline claim would likely require re-implementing the augmentation pipeline from the paper text alone.
Research summary
Jiangfeng Liu is an early-stage researcher in machine learning for energy systems, visible in the Evolve catalog through a single first-author arXiv preprint: 'Multi-Appliance Non-Intrusive Load Monitoring via Label-Preserving Aggregate Recomposition and Prediction Consistency' (2026). The paper addresses energy disaggregation — recovering individual appliance consumption from a household aggregate signal — with a training-only recipe: a residual-background swap that recomposes aggregate signals while preserving labels, combined with a prediction-consistency objective. The work also involves mixture-of-experts modeling, domain generalization, and data augmentation. Evolve's review finds the gains solid on the REDD benchmark but not significant on UK-DALE and REFIT, and notes the absence of ablation studies and released code, yielding a medium reproducibility readiness and a reproducibility score of 55/100. The visible record is narrow: exactly one reviewed paper, one distinct co-author (Yanfang Fan), first-author position without corresponding status, and a collaboration score of 14/100. Field concentration is maximal (100/100) simply because the sole reviewed paper sits in NILM / smart grids. Output, impact, and activity dimensions cannot be scored from the available data. No institution, country, or ORCID is on file, and 'Jiangfeng Liu' is a common name, so this profile is name-resolved from public bibliographic data and should be read as provisional until more papers are reviewed.
Collaboration pattern
The visible collaboration footprint is minimal: one reviewed paper with exactly one distinct co-author, Yanfang Fan, and a collaboration score of 14/100. Jiangfeng Liu is first author but not corresponding author on that work. With such a thin record, the single-partner pattern cannot yet be read as a deliberate solo style; it more likely reflects an early-stage profile where co-author networks have not accumulated.
Research keywords
Highlights
Training-only NILM via aggregate recomposition and prediction consistency
Benchmark gains concentrated on REDD
No ablations or released code yet
Timeline
- 2026Posts a first-author arXiv preprint on multi-appliance NILM via label-preserving aggregate recomposition and prediction consistency — the first reviewed work on record.
- 2026Evolve reviews the preprint: solid REDD gains, non-significant on UK-DALE and REFIT, no ablation or code; reproducibility readiness rated medium (55/100).
- 2026Collaborates with a single co-author, Yanfang Fan — the only distinct co-author in the current record, yielding a collaboration score of 14/100.
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.
- All bibliometric scores are computed from a single reviewed arXiv preprint, so every radar dimension is provisional.
- No institution, country, ORCID, or citation data is on file; 'Jiangfeng Liu' is a common name and the profile is name-resolved from public bibliographic data.
- Output, impact, and activity axes are unscored because the underlying data is insufficient.
- The reproducibility score averages one scored paper only (reproPapers = 1) and should not be generalized.