USC theorist spanning algorithmic game theory and learning theory, active on sample complexity and PAC learning
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
26
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—
The output axis is not scored yet: this radar lacks a validated publication-volume record for the profile. Once populated, it would measure yearly works, venue spread, and authorship roles across the 2008–2026 span.
Impact—
The impact axis is not scored yet: citation data has not been validated for this profile, even though the raw snapshot lists an h-index of 26 and an i10-index of 47. A completed score would measure citation influence.
Focus0/100
Focus scores 0/100, flagging an unusually broad spread rather than one narrow lane. The topic counts span five clusters — game theory and voting systems (83), auction theory and applications (59), game theory and applications (46), graph complexity and algorithms (43), and optimization and search problems (28) — while the recent reviewed papers sit squarely in machine learning theory, split evenly with statistical learning theory in the field distribution.
Collaboration22/100
Collaboration scores 22/100 on a small recent footprint: the radar records only 2 distinct co-authors, with work concentrated on Alireza F. Pour and Ganghua Wang. This points to compact, repeat partnerships rather than large rotating teams. The low score describes this recent, tightly concentrated pattern, not a measure of career-long collaboration breadth.
Activity90/100
Activity scores 90/100, among this profile's strongest signals. Per-year counts run from 2008 through 2026 with early peaks of 10 works (2009) and 11 works (2011 and 2015); after a quieter 2018–2023 stretch of two to four works per year, output rebounded to 8 works in 2025 and 9 in 2026, totalling 21 works in the last three years.
Reproducibility38/100
Reproducibility scores 38/100, computed from 2 Evolve-reviewed papers (reproMean 38). Readiness differs by paper: the membership-query active learning work is rated medium, while the instance-optimal learning paper is rated low, consistent with theory-first contributions whose claims rest on proofs rather than artifacts.
Research summary
Shaddin Dughmi is a researcher at the University of Southern California whose profile is resolved from OpenAlex bibliographic records (ORCID 0000-0002-2784-1868). The work has long roots in algorithmic game theory: topic counts are led by game theory and voting systems (83), auction theory and applications (59), game theory and applications (46), complexity and algorithms in graphs (43), and optimization and search problems (28). The two Evolve-reviewed 2026 papers, however, both fall in machine learning theory — one first-authored, one last-authored — which splits the field distribution evenly with statistical learning theory. The publication record spans 2008 through 2026, with annual counts peaking at 10 works in 2009 and 11 works in 2011 and again in 2015, a quieter 2018–2023 stretch of two to four works per year, and a rebound to 8 works in 2025 and 9 in 2026 — 21 works over the last three years. The snapshot lists an h-index of 26 and an i10-index of 47, although the radar leaves the output and impact axes unscored for now. Recent collaboration concentrates on Alireza F. Pour and Ganghua Wang.
Collaboration pattern
The current collaboration footprint is compact: the radar records only 2 distinct co-authors, and the profile's frequent co-authors are Alireza F. Pour and Ganghua Wang. Of the two 2026 reviewed papers, one is first-authored and one last-authored, with neither marked corresponding — a pattern consistent with the small, stable teams typical of theory research rather than large experimental collaborations.
Research keywords
Highlights
Resolving open questions in instance-optimal PAC learning
Membership queries make 2D halfspaces exponentially learnable
A two-decade arc from game theory to learning theory
Beyond the reviewed papers, aggregate topic counts show the deepest clusters are game theory and voting systems (83), auction theory and applications (59), and game theory and applications (46), with graph complexity and algorithms (43) and optimization and search problems (28) also prominent — the base from which the current machine-learning-theory line grew.
Timeline
- 2008Earliest year in the per-year publication counts, with 1 work recorded.
- 2009Early surge: 10 works recorded, among the highest annual totals of the record.
- 2011Peak productivity with 11 works; the same annual level recurs in 2015.
- 2018Start of a quieter stretch: annual counts of two to four works persist through 2023.
- 2026Resurgence with a learning-theory focus: 9 works recorded (8 in 2025), including two Evolve-reviewed arXiv papers — a first-authored instance-optimal learning paper and a last-authored active learning paper.
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 drawn from an OpenAlex-resolved public snapshot (identity: openalex) and were not independently verified against publisher records.
- The radar flags the output and impact axes as insufficient; h-index (26) and i10-index (47) appear in the raw snapshot but no output or impact scores are claimed here.
- Qualitative judgments rest on only two Evolve-reviewed papers, both 2026 arXiv preprints; the longer game-theory and auction-theory record is reflected solely in aggregate topic counts.
- Reproducibility (reproMean 38) is computed from just 2 papers and may shift as more work is reviewed.