Arun Kumar A

—

Unclaimed · AI-structured profileOpenAlex

Machine learning researcher spanning personalized federated learning and sample-efficient imitation learning

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) 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: beyond three 2026 arXiv preprints, the pipeline holds no verifiable publication-volume record for this name. This axis would count total works, venues, and annual output once bibliographic data is confirmed.

Impact—

Not yet scored: citation influence cannot be assessed — the snapshot reports h-index 0 and i10-index 0 with no citations by year. This axis would measure citation counts, h-index trajectory, and citing venues.

Focus0/100

Score 0 reflects maximal topical spread in this record: the OpenAlex field distribution splits evenly — machine learning (federated learning) 0.5 and robot learning / imitation learning 0.5 — and five distinct topics appear, led by Domain Adaptation and Few-Shot Learning (2 works) with single entries in robotics, privacy-preserving technologies, and healthcare AI. The profile reads as exploratory rather than concentrated, which is typical at this career stage.

Collaboration55/100

Score 55, built on 15 distinct co-authors in the recent window. Authorship roles vary: first author on SAPE-FL (federated learning), middle author on DISEIL (imitation learning) and SMOTE-VAR. Recurring partners include Sunil Gupta, Svetha Venkatesh, Dang Nguyen, and Bao Duong, while names such as Jill Newby and Helen Christensen in the co-author list align thematically with the depression-remission study.

Activity40/100

Score 40. All recorded output clusters in 2026: counts-by-year lists 3 works, matching recent3yWorks = 3, with no earlier years present in the snapshot. Three arXiv preprints in a single year is a concentrated burst; judging sustained research pace would require a longer observation window.

Reproducibility55/100

Score 55 (reproMean), averaged over the 2 papers Evolve has reviewed. Both carry medium repro-readiness: DISEIL wins across 10 simulation settings but is simulation-only with a privileged-state descriptor, and SAPE-FL shows broad empirical gains alongside idealized convergence theory — solid empirical practice with documented limitations rather than fully turnkey replication.

Research summary

Arun Kumar A is an early-career machine learning researcher whose profile is name-resolved from OpenAlex bibliographic data; no institution, country, or ORCID is attached to this record. The visible output consists of three 2026 arXiv preprints, and the OpenAlex field distribution splits evenly: machine learning (federated learning) at 0.5 and robot learning / imitation learning at 0.5. The first-author paper, SAPE-FL, proposes similarity-aware personalized federated learning for heterogeneous, non-IID environments; Evolve's review notes broad empirical gains but idealized convergence theory. As a middle author on DISEIL, the researcher contributes to demonstration distillation for sample-efficient imitation learning, a method that topped all 10 simulated settings by modest margins while remaining simulation-only. A third paper, SMOTE-VAR, applies uncertainty-aware oversampling to predicting depression remission in university students and has not yet been reviewed by Evolve. OpenAlex topics are led by Domain Adaptation and Few-Shot Learning (2 works), with single entries in reinforcement learning in robotics, robot manipulation, privacy-preserving technologies, and healthcare AI. The snapshot reports h-index 0 and i10-index 0, 3 works in the recent three-year window, and 15 distinct co-authors — an exploratory, output-stage profile best read together with its caveats.

Collaboration pattern

The record lists 15 distinct co-authors across the recent window. Frequent partners include Sunil Gupta, Svetha Venkatesh, Dang Nguyen, and Bao Duong on the machine learning side, alongside clinically oriented names such as Jill Newby, Helen Christensen, Alexis Whitton, Taylor A. Braund, Leonard Hoon, Wu Yi Zheng, Debopriyo Bal, and Leonard Hoon's co-listed mental-health colleagues. Authorship alternates between first author (SAPE-FL) and middle positions (DISEIL, SMOTE-VAR), indicating both independent and contributory collaboration modes.

Research keywords

federated learningpersonalized federated learningnon-IID dataclient clusteringinteractive imitation learningDAggerdemonstration efficiencyfailure-mode clusteringvision-language modelsrobot manipulationfew-shot learningAI in healthcare

Highlights

First-author SAPE-FL for personalized federated learning
First-author 2026 arXiv preprint anchoring clients to global and peer models via hybrid similarity to handle statistical heterogeneity and non-IID data; claims state of the art in personalized FL, with broad empirical gains but convergence theory that Evolve flags as idealized.
DISEIL distills demonstrations for sample-efficient imitation learning
Middle-author 2026 arXiv preprint that requests demonstrations for chosen failure modes and start states, clustering failures to pick demos efficiently; it tops all 10 simulated settings (1 tie) by modest margins, though it is simulation-only with a privileged-state descriptor.
Uncertainty-aware oversampling for depression-remission prediction
SMOTE-VAR (2026, middle author) targets class imbalance in predicting depression remission among university students using an uncertainty-aware oversampling method; this paper has not yet been reviewed by Evolve, so no verdict, keywords, or repro-readiness exist for it.

Timeline

2026First-author arXiv preprint of SAPE-FL, similarity-aware personalized federated learning in heterogeneous environments
2026Middle-author arXiv preprint of DISEIL, demonstration distillation for sample-efficient imitation learning
2026Middle-author arXiv preprint of SMOTE-VAR, uncertainty-aware oversampling for predicting depression remission in university students
2026OpenAlex counts-by-year records 3 works in 2026, the only publication year in this snapshot

Papers in the Evolve catalog

2026-09-08
DISEIL: Demonstration Distillation for Sample-Efficient Imitation Learning
2026-09-02first author
Similarity-Aware Personalized Federated Learning in Heterogeneous Environments

Frequent co-authors

Sunil Gupta×2 Dang Nguyen×1 Taylor A. Braund×1 Wu Yi Zheng×1 Debopriyo Bal×1 Leonard Hoon×1 Jill Newby×1 Helen Christensen×1 Svetha Venkatesh×1 Alexis Whitton×1 Dang Ngyuen×1 Bao Duong×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 metrics derive from OpenAlex only; the snapshot reports h-index 0 and i10-index 0 with no citation history, so impact is not yet measurable.
No institution, country, or ORCID is available; the profile is name-resolved from public bibliographic data and 'Arun Kumar A' is a common name pattern, so identity should be verified before use.
The scope is limited to three works, all 2026 arXiv preprints; no peer-reviewed venue record or publication history before 2026 is included.
SMOTE-VAR has not been reviewed by Evolve, so no verdict, field, keywords, or repro-readiness exist for it.