Jun Zhou

—

Unclaimed · AI-structured profileOpenAlex

ML researcher spanning brain-inspired continual learning and LLM agent evaluation for financial information retrieval

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—

This axis measures publication volume and productivity over time. The underlying counting data for this profile is not yet available, so output is honestly left unscored rather than estimated from the two visible catalog papers.

Impact—

This axis measures citation-based influence — citation counts, aggregated indices, and uptake of the work by other researchers. No citation metrics are present in this snapshot, so the dimension is left unscored.

Focus0/100

Focus scores 0, reflecting an exactly even field distribution: 0.5 machine learning (NeuroAI/continual learning) and 0.5 natural language processing (LLM agent evaluation). The two 2026 papers sit one per field — one brain-inspired continual learning study and one financial retrieval benchmark — so this reads as a genuine two-field profile rather than a data gap.

Collaboration64/100

Collaboration scores 64, grounded in 24 distinct coauthors recorded across the profile. The frequent-coauthor list is led by repeat entries Hongwei Yan and Kanglei Zhou, alongside Qi Cheng, Weiyi Dong, Chunyan Lan, Guanglong Sun, Qian Li, Yi Zhong, and Liyuan Wang. Both 2026 papers are multi-author efforts, with Jun Zhou in the last position on FinFIRST and a middle position on the continual learning paper.

Activity—

This axis measures recent publishing cadence and year-over-year consistency of research output. Only two 2026 papers are visible in this snapshot, with no multi-year publication history attached, so activity is left unscored.

Reproducibility55/100

Reproducibility scores 55, computed as a mean across 2 assessed papers, each rated medium reproducibility readiness. For FinFIRST, Evolve notes rigorous benchmark construction while cautioning that single-run LLM-judged results warrant care; the brain-inspired continual learning paper is judged strong empirically but with informal theory and replay-competitive results. With only two papers, treat this as an early signal.

Research summary

This profile is name-resolved from public bibliographic data for Jun Zhou; "Jun Zhou" is a common name, and no institutional affiliation, country, or ORCID is attached to this record. The Evolve catalog contains two 2026 arXiv papers. The first, FinFIRST, lists Jun Zhou as a last-position author on a benchmark that scores LLM search agents on financial information retrieval, sourcing, and traceability through atomic rubrics; Evolve judges the construction rigorous but cautions that single-run LLM-judged results warrant care. The second, Brain-Inspired Hierarchical Modularity for General Continual Learning, is a middle-author contribution applying Drosophila mushroom-body-inspired mixture-of-experts and ensemble modularity to continual learning across vision and robotics. The field split is exactly even: 0.5 machine learning (NeuroAI/continual learning) and 0.5 natural language processing (LLM agent evaluation), yielding a focus score of 0 — a genuine two-field profile rather than a single-field one. The collaboration dimension scores 64, with 24 distinct coauthors recorded; Hongwei Yan and Kanglei Zhou are the repeat entries. Reproducibility scores 55 as a mean across 2 assessed papers, both rated medium readiness. Output, impact, and activity dimensions lack sufficient underlying data and are left unscored. No citation counts, h-index, or venue-level statistics are available in this snapshot.

Collaboration pattern

The snapshot records 24 distinct coauthors, and the collaboration dimension scores 64. Hongwei Yan and Kanglei Zhou are the only repeat entries in the frequent-coauthor list, alongside Qi Cheng, Weiyi Dong, Chunyan Lan, Guanglong Sun, Qian Li, Yi Zhong, and Liyuan Wang. Both 2026 papers are team efforts in which Jun Zhou holds a last position on FinFIRST and a middle position on the continual learning paper, consistent with contributing within larger multi-author groups.

Research keywords

continual learningcatastrophic forgettingmixture-of-expertsensemble learningDrosophila mushroom bodyparameter-efficient fine-tuningfinancial information retrievalLLM search agentsbenchmark constructionatomic rubric evaluationLLM-as-judgeevidence traceability

Highlights

FinFIRST: jointly scoring answers and evidence for financial search agents
As a last-position author, Jun Zhou contributes to a benchmark that evaluates LLM search agents on financial information retrieval, sourcing, and traceability via atomic rubrics. Evolve flags the construction as rigorous but cautions that single-run LLM-judged results warrant caution.
Brain-inspired hierarchical modularity for general continual learning
In a middle-author role, Jun Zhou contributes to a Drosophila mushroom-body-inspired design that combines mixture-of-experts and ensemble modularity, yielding large continual-learning gains across vision and robotics. Evolve notes strong empirics alongside informal theory and replay-competitive results.
Modular continual learning that stays replay-competitive
Evolve's verdict on the brain-inspired paper records that the Drosophila-inspired MoE-plus-ensemble modularity is replay-competitive, with strong empirical gains in vision and robotics despite an informal theoretical grounding — an argument for architectural, biology-derived anti-forgetting mechanisms.

Timeline

2026Last-position author on FinFIRST (arXiv), benchmarking LLM search agents for financial information retrieval, sourcing, and traceability.
2026Middle-author contribution to Brain-Inspired Hierarchical Modularity for General Continual Learning (arXiv), applying Drosophila-inspired mixture-of-experts and ensemble modularity.
2026Both cataloged papers receive medium reproducibility readiness, with a reproducibility mean of 55 across the 2 assessed papers.
2026Bibliometric snapshot records 24 distinct coauthors and an even 0.5/0.5 field split between machine learning and natural language processing.

Papers in the Evolve catalog

Frequent co-authors

Hongwei Yan×1 Kanglei Zhou×1 Qi Cheng×1 Weiyi Dong×1 Chunyan Lan×1 Guanglong Sun×1 Qian Li×1 Yi Zhong×1 Liyuan Wang×1 Jun Zhou×1 Hongwei Yan×1 Kanglei Zhou×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 dimensions are computed from OpenAlex-derived public bibliometric data; no institutional affiliation, country, or ORCID is attached to this record.
"Jun Zhou" is a common name; this profile is name-resolved from public bibliographic data and may merge or split identities.
Output, impact, and activity dimensions lack sufficient underlying data and are left unscored; only 2 papers and 24 distinct coauthors are visible in this snapshot.
Reproducibility (55) is a mean over only 2 assessed papers, both rated medium readiness, so it is a small-sample signal.