FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912592113> ?p ?o ?g. }

• W2912592113 abstract "A key learning scenario in large-scale applications is that of federated learning, where a centralized model is trained based on data originating from a large number of clients. We argue that, with the existing training and inference, federated models can be biased towards different clients. Instead, we propose a new framework of agnostic federated learning, where the centralized model is optimized for any target distribution formed by a mixture of the client distributions. We further show that this framework naturally yields a notion of fairness. We present data-dependent Rademacher complexity guarantees for learning with this objective, which guide the definition of an algorithm for agnostic federated learning. We also give a fast stochastic optimization algorithm for solving the corresponding optimization problem, for which we prove convergence bounds, assuming a convex loss function and hypothesis set. We further empirically demonstrate the benefits of our approach in several datasets. Beyond federated learning, our framework and algorithm can be of interest to other learning scenarios such as cloud computing, domain adaptation, drifting, and other contexts where the training and test distributions do not coincide." @default.
• W2912592113 created "2019-02-21" @default.
• W2912592113 creator A5058849006 @default.
• W2912592113 creator A5062844614 @default.
• W2912592113 creator A5085885345 @default.
• W2912592113 date "2019-02-01" @default.
• W2912592113 modified "2023-10-01" @default.
• W2912592113 title "Agnostic Federated Learning" @default.
• W2912592113 cites W1508165687 @default.
• W2912592113 cites W1632114991 @default.
• W2912592113 cites W1722318740 @default.
• W2912592113 cites W1822439997 @default.
• W2912592113 cites W1856231595 @default.
• W2912592113 cites W1978920452 @default.
• W2912592113 cites W2048087720 @default.
• W2912592113 cites W2049901611 @default.
• W2912592113 cites W2096765209 @default.
• W2912592113 cites W2100969003 @default.
• W2912592113 cites W2101460669 @default.
• W2912592113 cites W2102605133 @default.
• W2912592113 cites W2107369107 @default.
• W2912592113 cites W2111362445 @default.
• W2912592113 cites W2125320996 @default.
• W2912592113 cites W2146871184 @default.
• W2912592113 cites W2157989183 @default.
• W2912592113 cites W2158815628 @default.
• W2912592113 cites W2163302275 @default.
• W2912592113 cites W2288408124 @default.
• W2912592113 cites W2530417694 @default.
• W2912592113 cites W2535838896 @default.
• W2912592113 cites W2547352193 @default.
• W2912592113 cites W2618494520 @default.
• W2912592113 cites W2750384547 @default.
• W2912592113 cites W2765094320 @default.
• W2912592113 cites W2810084418 @default.
• W2912592113 cites W2900120080 @default.
• W2912592113 cites W2949436635 @default.
• W2912592113 cites W2950538796 @default.
• W2912592113 cites W2950539955 @default.
• W2912592113 cites W2951670162 @default.
• W2912592113 cites W2953226914 @default.
• W2912592113 cites W2953362272 @default.
• W2912592113 cites W2963189173 @default.
• W2912592113 cites W2963826681 @default.
• W2912592113 cites W879220392 @default.
• W2912592113 cites W96659543 @default.
• W2912592113 hasPublicationYear "2019" @default.
• W2912592113 type Work @default.
• W2912592113 sameAs 2912592113 @default.
• W2912592113 citedByCount "39" @default.
• W2912592113 countsByYear W29125921132019 @default.
• W2912592113 countsByYear W29125921132020 @default.
• W2912592113 countsByYear W29125921132021 @default.
• W2912592113 crossrefType "posted-content" @default.
• W2912592113 hasAuthorship W2912592113A5058849006 @default.
• W2912592113 hasAuthorship W2912592113A5062844614 @default.
• W2912592113 hasAuthorship W2912592113A5085885345 @default.
• W2912592113 hasConcept C111919701 @default.
• W2912592113 hasConcept C119857082 @default.
• W2912592113 hasConcept C120665830 @default.
• W2912592113 hasConcept C121332964 @default.
• W2912592113 hasConcept C134306372 @default.
• W2912592113 hasConcept C139807058 @default.
• W2912592113 hasConcept C14036430 @default.
• W2912592113 hasConcept C154945302 @default.
• W2912592113 hasConcept C162324750 @default.
• W2912592113 hasConcept C177264268 @default.
• W2912592113 hasConcept C199360897 @default.
• W2912592113 hasConcept C26517878 @default.
• W2912592113 hasConcept C2776214188 @default.
• W2912592113 hasConcept C2777303404 @default.
• W2912592113 hasConcept C2992525071 @default.
• W2912592113 hasConcept C33923547 @default.
• W2912592113 hasConcept C36503486 @default.
• W2912592113 hasConcept C38652104 @default.
• W2912592113 hasConcept C41008148 @default.
• W2912592113 hasConcept C50522688 @default.
• W2912592113 hasConcept C78458016 @default.
• W2912592113 hasConcept C79974875 @default.
• W2912592113 hasConcept C80444323 @default.
• W2912592113 hasConcept C86803240 @default.
• W2912592113 hasConceptScore W2912592113C111919701 @default.
• W2912592113 hasConceptScore W2912592113C119857082 @default.
• W2912592113 hasConceptScore W2912592113C120665830 @default.
• W2912592113 hasConceptScore W2912592113C121332964 @default.
• W2912592113 hasConceptScore W2912592113C134306372 @default.
• W2912592113 hasConceptScore W2912592113C139807058 @default.
• W2912592113 hasConceptScore W2912592113C14036430 @default.
• W2912592113 hasConceptScore W2912592113C154945302 @default.
• W2912592113 hasConceptScore W2912592113C162324750 @default.
• W2912592113 hasConceptScore W2912592113C177264268 @default.
• W2912592113 hasConceptScore W2912592113C199360897 @default.
• W2912592113 hasConceptScore W2912592113C26517878 @default.
• W2912592113 hasConceptScore W2912592113C2776214188 @default.
• W2912592113 hasConceptScore W2912592113C2777303404 @default.
• W2912592113 hasConceptScore W2912592113C2992525071 @default.
• W2912592113 hasConceptScore W2912592113C33923547 @default.
• W2912592113 hasConceptScore W2912592113C36503486 @default.
• W2912592113 hasConceptScore W2912592113C38652104 @default.
• W2912592113 hasConceptScore W2912592113C41008148 @default.

• next