Concave Statistical Utility Maximization Bandits via Influence-Function Gradients
2026-04-27T07:24:07Z•1ef7dfb240767cadeef148f4055605d351c5e4d26c9eed1fce3a8cdd5d997b7a
Nystrom approximationProjAvgRLAvgSPD matricesSQUEAKStiefel manifoldSuper LearnerWassersteinGradconcave utilityconformal predictionconformalized Super Learnerentropic Wasserstein barycenter','explainability','feature at-federated learninggeometry-aware aggregationinfluence functionkernel density/proposal learningkernel ridge regressionmirror ascentmulti-armed banditsmultiplicative weightspliable rejection samplingrejection samplingridge leverage scoresstochastic gradientsunnormalized RLS
What happened
Collection of new stat-ML papers (arXiv 2026-04-27) introducing methods across bandits, sampling, kernel approximation, ensemble uncertainty, federated geometry-aware networks, explainability for dynamic fields, mixed-membership clustering, customer-revenue forecasting, and audits of LLM probabilistic sampling. Highlights include: (1) concave-utility multi-armed bandits optimized via influence-function gradients and entropic mirror-ascent; (2) pliable rejection sampling (PRS) learning proposals with kernel estimators and acceptance guarantees; (3) SQUEAK, an efficient unnormalized RLS-based Ny
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_stat_ml
- Record identifier
- 1ef7dfb240767cadeef148f4055605d351c5e4d26c9eed1fce3a8cdd5d997b7a
- Enrichment time
- 2026-04-27T07:24:07Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.