Hong Shen

Central Queensland University · Australia

Unclaimed · AI-structured profileOpenAlex

Federated learning security researcher: Byzantine-robust decentralized learning and backdoor attacks

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-16) 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 total publication volume and cumulative metrics such as the h-index and i10-index. The current OpenAlex record shows h-index 0 and i10-index 0 with no verified work list, so output cannot be scored here and no estimate is made.

Impact—

Impact would track citation counts, citation trajectory over time, and uptake of the reviewed papers. No citation data is available in the current snapshot — h-index and i10-index are both recorded as 0 — so this axis is left unscored rather than estimated.

Focus0/100

The focus score of 0 reflects a spread record: the OpenAlex field distribution splits evenly, one third each across distributed machine learning (federated learning), federated learning and distributed optimization, and machine learning security; topic assignments span five clusters — privacy-preserving technologies (4), domain adaptation and few-shot learning (2), adversarial robustness (2), cryptography and data security (1), stochastic gradient optimization (1). The themes are adjacent but not yet concentrated on a single line.

Collaboration41/100

The collaboration score of 41 rests on seven distinct co-authors across four recent works. The reviewed papers show a recurring team — Xiao Ma, Hui Tian, Wei Ke, Wenqi Lyu, Jian Wang and Xue Hua Liu — but on all three papers Shen holds a middle author position and is never the corresponding author, indicating contribution within a group rather than leading one under the current record.

Activity47/100

Activity is scored 47 on thin but real evidence: recent3yWorks stands at 4, and the counts-by-year series contains a single entry — 4 works in 2026. All three Evolve-reviewed papers are 2026 arXiv preprints, so the visible output clusters entirely in one year with no earlier history in this record, consistent with an early-stage or newly-resolved profile rather than an established publication cadence.

Reproducibility43/100

Reproducibility averages 43 across three assessed papers: two rated medium readiness — robust decentralized personalized federated learning and robust decentralized federated distillation — and one rated low, the fine-grained distributed backdoor attack. Evolve's verdicts flag single-dataset evaluation with thin baselines, roughly 1-point gains over non-robust baselines in single runs, and candid but stylized attack experiments without significance tests.

Research summary

Hong Shen is a researcher at Central Queensland University in Australia whose recent work concentrates on the security and robustness of federated learning systems. OpenAlex bibliographic data resolves the profile into three evenly weighted fields — distributed machine learning (federated learning), federated learning and distributed optimization, and machine learning security — each accounting for one third of the classified record. All four works recorded in the last three years fall in 2026, and all three Evolve-reviewed papers are arXiv preprints from that year. Together they form a coherent programme: Byzantine-robust decentralized personalized federated learning, Byzantine-robust decentralized federated distillation for heterogeneous models, and a fine-grained distributed backdoor attack reported to need 37–48% less poisoning than DBA. Across these papers Shen appears as a middle-position author and never as corresponding author, working with seven distinct co-authors including Xiao Ma, Hui Tian and Wei Ke. Evolve's reproducibility assessment averages 43 across three papers — two medium, one low — with recurring criticism of single-dataset evaluation, thin or non-robust baselines, and single runs without significance testing.

Collaboration pattern

Across the three Evolve-reviewed 2026 papers, Hong Shen appears as a middle-position author and never as corresponding author, working within one recurring team. Seven distinct co-authors appear in the recent record, with Xiao Ma, Hui Tian, Wei Ke, Wenqi Lyu, Jian Wang and Xue Hua Liu listed among frequent collaborators. The pattern suggests a consistent small-group collaboration around federated learning robustness rather than an independent group lead — based solely on this snapshot.

Research keywords

federated learningdecentralized federated learningpersonalized federated learningByzantine robustnessrobust aggregationknowledge distillationmodel heterogeneitynon-IID databackdoor attackdata poisoningadversarial robustnessadaptive clipping

Highlights

Byzantine-robust decentralized personalized federated learning (R-DPFL)
Evolve's review finds R-DPFL defends personalized decentralized federated learning against Byzantine attacks through robust aggregation, prediction and clipping, while noting that evaluation rests on a single dataset with thin baselines.
Byzantine-robust decentralized federated distillation across heterogeneous models
The reviewed distillation work couples credible theory with Byzantine-robust aggregation for model-heterogeneous, non-IID settings, but Evolve records gains of about 1 point over non-robust baselines in single runs.
Fine-grained distributed backdoor attack on federated learning
FDBA proposes an edge-trigger distributed backdoor with dynamic trigger generation, reported to require 37–48% less poisoning than DBA; Evolve credits broad, candid experiments but flags stylized theory and no significance tests.

Timeline

2026Publishes Fine-grained Distributed Backdoor Attacks in Federated Learning on arXiv, proposing dynamic trigger generation with 37–48% less poisoning than DBA.
2026Publishes Robust Decentralized Federated Distillation via Multi-Modality Knowledge Collaboration on arXiv, extending Byzantine robustness to model-heterogeneous federated distillation.
2026Publishes Robust Decentralized Personalized Federated Learning on arXiv, adding prediction-constrained neighborhood collaboration to Byzantine-robust personalization.
2026All 4 works recorded in the OpenAlex counts-by-year data fall in 2026, produced with 7 distinct co-authors across federated learning security and robustness topics.

Papers in the Evolve catalog

Frequent co-authors

Xiao Ma×2 Hui Tian×2 Wei Ke×2 Wenqi Lyu×2 Wei Ke×1 Jian Wang×1 Xue Hua Liu×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 are computed from OpenAlex data only; h-index and i10-index read 0, and the output and impact axes are unscored due to insufficient data.
All three catalog papers are 2026 arXiv preprints; peer-reviewed publication status is not captured in this snapshot.
"Hong Shen" is a common name; this profile is name-resolved from public bibliographic data, and no ORCID was available for disambiguation.
The activity and collaboration picture rests on only 4 recent works and 7 distinct co-authors; earlier career history falls outside this record.