Matches in SemOpenAlex for { <https://semopenalex.org/work/W4320461957> ?p ?o ?g. }
Showing items 1 to 67 of
67
with 100 items per page.
- W4320461957 abstract "We consider prediction with expert advice for strongly convex and bounded losses, and investigate trade-offs between regret and variance (i.e., squared difference of learner's predictions and best expert predictions). With $K$ experts, the Exponentially Weighted Average (EWA) algorithm is known to achieve $O(log K)$ regret. We prove that a variant of EWA either achieves a negative regret (i.e., the algorithm outperforms the best expert), or guarantees a $O(log K)$ bound on both variance and regret. Building on this result, we show several examples of how variance of predictions can be exploited in learning. In the online to batch analysis, we show that a large empirical variance allows to stop the online to batch conversion early and outperform the risk of the best predictor in the class. We also recover the optimal rate of model selection aggregation when we do not consider early stopping. In online prediction with corrupted losses, we show that the effect of corruption on the regret can be compensated by a large variance. In online selective sampling, we design an algorithm that samples less when the variance is large, while guaranteeing the optimal regret bound in expectation. In online learning with abstention, we use a similar term as the variance to derive the first high-probability $O(log K)$ regret bound in this setting. Finally, we extend our results to the setting of online linear regression." @default.
- W4320461957 created "2023-02-14" @default.
- W4320461957 creator A5037592528 @default.
- W4320461957 creator A5071419221 @default.
- W4320461957 creator A5074621968 @default.
- W4320461957 date "2022-06-06" @default.
- W4320461957 modified "2023-09-29" @default.
- W4320461957 title "A Regret-Variance Trade-Off in Online Learning" @default.
- W4320461957 doi "https://doi.org/10.48550/arxiv.2206.02656" @default.
- W4320461957 hasPublicationYear "2022" @default.
- W4320461957 type Work @default.
- W4320461957 citedByCount "0" @default.
- W4320461957 crossrefType "posted-content" @default.
- W4320461957 hasAuthorship W4320461957A5037592528 @default.
- W4320461957 hasAuthorship W4320461957A5071419221 @default.
- W4320461957 hasAuthorship W4320461957A5074621968 @default.
- W4320461957 hasBestOaLocation W43204619571 @default.
- W4320461957 hasConcept C105795698 @default.
- W4320461957 hasConcept C11413529 @default.
- W4320461957 hasConcept C119857082 @default.
- W4320461957 hasConcept C121955636 @default.
- W4320461957 hasConcept C126255220 @default.
- W4320461957 hasConcept C134306372 @default.
- W4320461957 hasConcept C136764020 @default.
- W4320461957 hasConcept C149782125 @default.
- W4320461957 hasConcept C162324750 @default.
- W4320461957 hasConcept C196083921 @default.
- W4320461957 hasConcept C196921405 @default.
- W4320461957 hasConcept C2986087404 @default.
- W4320461957 hasConcept C33923547 @default.
- W4320461957 hasConcept C34388435 @default.
- W4320461957 hasConcept C41008148 @default.
- W4320461957 hasConcept C50817715 @default.
- W4320461957 hasConcept C77553402 @default.
- W4320461957 hasConceptScore W4320461957C105795698 @default.
- W4320461957 hasConceptScore W4320461957C11413529 @default.
- W4320461957 hasConceptScore W4320461957C119857082 @default.
- W4320461957 hasConceptScore W4320461957C121955636 @default.
- W4320461957 hasConceptScore W4320461957C126255220 @default.
- W4320461957 hasConceptScore W4320461957C134306372 @default.
- W4320461957 hasConceptScore W4320461957C136764020 @default.
- W4320461957 hasConceptScore W4320461957C149782125 @default.
- W4320461957 hasConceptScore W4320461957C162324750 @default.
- W4320461957 hasConceptScore W4320461957C196083921 @default.
- W4320461957 hasConceptScore W4320461957C196921405 @default.
- W4320461957 hasConceptScore W4320461957C2986087404 @default.
- W4320461957 hasConceptScore W4320461957C33923547 @default.
- W4320461957 hasConceptScore W4320461957C34388435 @default.
- W4320461957 hasConceptScore W4320461957C41008148 @default.
- W4320461957 hasConceptScore W4320461957C50817715 @default.
- W4320461957 hasConceptScore W4320461957C77553402 @default.
- W4320461957 hasLocation W43204619571 @default.
- W4320461957 hasOpenAccess W4320461957 @default.
- W4320461957 hasPrimaryLocation W43204619571 @default.
- W4320461957 hasRelatedWork W2139129601 @default.
- W4320461957 hasRelatedWork W2186411897 @default.
- W4320461957 hasRelatedWork W2609240131 @default.
- W4320461957 hasRelatedWork W2998428385 @default.
- W4320461957 hasRelatedWork W3149052698 @default.
- W4320461957 hasRelatedWork W3176910946 @default.
- W4320461957 hasRelatedWork W3211820441 @default.
- W4320461957 hasRelatedWork W4225127995 @default.
- W4320461957 hasRelatedWork W4282979967 @default.
- W4320461957 hasRelatedWork W4304208721 @default.
- W4320461957 isParatext "false" @default.
- W4320461957 isRetracted "false" @default.
- W4320461957 workType "article" @default.