Learning theorist at NC State: finite-time guarantees for federated RL and stochastic optimization.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
17
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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 scored yet: the underlying snapshot lacks the publication-volume records this axis needs. Output would measure the number, venue quality, and year-over-year consistency of the researcher's published works.
Impact—
Not scored yet: citation-count and influence records are missing from the current snapshot, so impact cannot be rated. This axis would measure citation accumulation, uptake of the proposed methods, and the scholarly reach of the work.
Focus0/100
Scores 0/100: the OpenAlex field split is exactly 50% reinforcement learning theory and 50% stochastic optimization and learning theory, the most dispersed profile possible on this axis. Topic records reinforce that breadth — distributed control of multi-agent systems (35), stability and control of uncertain systems (21), reinforcement learning in robotics (21), privacy-preserving technologies in data (18), and stochastic gradient optimization techniques (17) — so the work spans several adjacent theory communities rather than one narrow niche.
Collaboration28/100
Scores 28/100, reflecting a deliberately compact collaboration footprint: only 3 distinct co-authors appear across 27 works in the recent three-year window, so the low score tracks co-author diversity rather than output. The recurring partners — Sreejeet Maity, Feng Zhu, and Robert W. Heath — carry much of that activity, and both 2026 arXiv preprints in the Evolve catalog place Mitra in the last author position, consistent with a stable senior role in a small team.
Activity97/100
Scores 97/100, near the top of the axis. The year-by-year record shows steady single-digit to low-double-digit output from 2018 (7 works) onward, a first peak of 10 works in 2021, and a clear acceleration to 15 works in 2024 — the highest year on record. With 5 works indexed in 2025 and 7 already in 2026, the recent three-year window holds 27 works, indicating a sustained and currently accelerating research tempo.
Reproducibility55/100
Scores 55/100, a mean built from 2 reviewed papers (reproducibility mean 55). Both carry a medium repro-readiness rating, and Evolve's verdicts explain why: the federated Q-learning proofs look sound but the experiments are thin, while the batched SGD analysis delivers short, consistent high-probability proofs with no experiments at all. Rigorous theory with limited empirical artifacts currently defines this axis.
Research summary
Aritra Mitra is a researcher at North Carolina State University working at the intersection of reinforcement learning theory and stochastic optimization, with a documented OpenAlex profile (h-index 17, i10-index 21). The bibliometric record shows 27 works over the most recent three-year window and a pronounced activity upswing: yearly output climbed from single digits through 2018-2023 to 15 works in 2024 and 7 indexed so far in 2026. Topic distribution points to distributed control of multi-agent systems (35 records) as the largest cluster, alongside stability and control of uncertain systems (21), reinforcement learning in robotics (21), privacy-preserving technologies (18), and stochastic gradient optimization (17). The two papers reviewed by Evolve — both 2026 arXiv preprints where Mitra is last author — reflect the current 50/50 field split between reinforcement learning theory and stochastic optimization and learning theory: a robust federated Q-learning scheme with median-of-means aggregation in Õ(1) communication rounds, and a high-probability last-iterate convergence analysis of batched SGD. Evolve's reviews credit sound, consistent proofs while flagging thin experimental support and, in one case, a T→∞ claim that clashes with a feasibility condition. Frequent co-authors include Sreejeet Maity, Feng Zhu, and Robert W. Heath.
Collaboration pattern
The collaboration footprint is compact: 3 distinct co-authors across 27 works in the recent three-year window, and a collaboration score of 28/100 driven by that narrow diversity. Three recurring partners dominate — Sreejeet Maity, Feng Zhu, and Robert W. Heath. In both 2026 arXiv preprints reviewed by Evolve, Mitra sits in the last author position (non-corresponding), consistent with a senior supervisory role within a small, stable theory group rather than large-scale cross-institution projects.
Research keywords
Highlights
Robust Federated Q-Learning in Õ(1) Communication Rounds
High-Probability Last-Iterate Convergence for Batched SGD
2024 Output Peak: 15 Indexed Works
The OpenAlex year-by-year record shows the busiest year of this profile is 2024, with 15 works — up from 10 in 2021 and 6 in 2023 — followed by 5 works in 2025 and 7 already indexed in 2026, for 27 works across the recent three-year window.
Timeline
- 2018Earliest sustained climb in the OpenAlex record: 7 indexed works in 2018, up from 1 in 2017.
- 2021First output peak of the record, with 10 indexed works in a single year.
- 2024Highest recorded output of the profile: 15 indexed works in one year, versus 6 in 2023.
- 2026Two arXiv preprints enter the Evolve catalog — robust federated Q-learning and high-probability batched SGD — in a year with 7 works already indexed.
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.
- Bibliometrics are from OpenAlex only; citation and venue-quality fields were not available in this snapshot, so output and impact are unscored.
- The profile is name-resolved from public bibliographic data (OpenAlex identity, ORCID 0000-0002-0472-284X); no biographical information beyond affiliation was verified.
- Reproducibility rests on only 2 Evolve-reviewed papers (mean 55); the wider corpus has not been reviewed.
- Yearly counts include partial years (2025: 5; 2026: 7 as of the snapshot), so recent-year totals are not final.