PFN-TS converts PFN posterior predictives into mean-reward samples for Thompson sampling using a subsampled predictive CLT, with consistency proofs, regret bounds, and strong empirical performance on synthetic and real bandit benchmarks.
ISBN 978-1-4503-0493-1
5 Pith papers cite this work, alongside 25 external citations. Polarity classification is still indexing.
5
Pith papers citing it
25
external citations · external index
citation-role summary
background 1
method 1
citation-polarity summary
representative citing papers
Eye-tracking study shows F-pattern and examination hypothesis from web search do not hold in carousel interfaces; users follow an L-pattern on clicks, ignore headings, and examination does not predict clicks as assumed.
A support-aware DSS integrates replay, OPE, lower-bound ranking, multi-sided guardrails, out-of-time validation, and interference-aware design to output launch-readiness classifications rather than single performance estimates, applied to RTB logs where a margin-gated floor policy is selected for va
Controlled personalization combining editorial curation with modest algorithmic recommendations in legacy news increases engagement, diversity, and reduces popularity bias per an A/B test.
citing papers explorer
-
PFN-TS: Thompson Sampling for Contextual Bandits via Prior-Data Fitted Networks
PFN-TS converts PFN posterior predictives into mean-reward samples for Thompson sampling using a subsampled predictive CLT, with consistency proofs, regret bounds, and strong empirical performance on synthetic and real bandit benchmarks.
-
Following the Eye-Tracking Evidence: Established Web-Search Assumptions Fail in Carousel Interfaces
Eye-tracking study shows F-pattern and examination hypothesis from web search do not hold in carousel interfaces; users follow an L-pattern on clicks, ignore headings, and examination does not predict clicks as assumed.
-
Decision Support for Marketplace Policies under Incomplete Evidence: From Replay to Launch Readiness
A support-aware DSS integrates replay, OPE, lower-bound ranking, multi-sided guardrails, out-of-time validation, and interference-aware design to output launch-readiness classifications rather than single performance estimates, applied to RTB logs where a margin-gated floor policy is selected for va
-
Controlled Personalization in Legacy Media Online Services: A Case Study in News Recommendation
Controlled personalization combining editorial curation with modest algorithmic recommendations in legacy news increases engagement, diversity, and reduces popularity bias per an A/B test.
- A More Accurate Algorithm Comparison through A/B Testing using Offline Evaluation Methods