FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3086152015> ?p ?o ?g. }

• W3086152015 abstract "Current federated learning algorithms take tens of communication rounds transmitting unwieldy model weights under ideal circumstances and hundreds when data is poorly distributed. Inspired by recent work on dataset distillation and distributed one-shot learning, we propose Distilled One-Shot Federated Learning (DOSFL) to significantly reduce the communication cost while achieving comparable performance. In just one round, each client distills their private dataset, sends the synthetic data (e.g. images or sentences) to the server, and collectively trains a global model. The distilled data look like noise and are only useful to the specific model weights, i.e., become useless after the model updates. With this weight-less and gradient-less design, the total communication cost of DOSFL is up to three orders of magnitude less than FedAvg while preserving between 93% to 99% performance of a centralized counterpart. Afterwards, clients could switch to traditional methods such as FedAvg to finetune the last few percent to fit personalized local models with local datasets. Through comprehensive experiments, we show the accuracy and communication performance of DOSFL on both vision and language tasks with different models including CNN, LSTM, Transformer, etc. We demonstrate that an eavesdropping attacker cannot properly train a good model using the leaked distilled data, without knowing the initial model weights. DOSFL serves as an inexpensive method to quickly converge on a performant pre-trained model with less than 0.1% communication cost of traditional methods." @default.
• W3086152015 created "2020-09-21" @default.
• W3086152015 creator A5022526821 @default.
• W3086152015 creator A5040043382 @default.
• W3086152015 creator A5061233361 @default.
• W3086152015 creator A5061695845 @default.
• W3086152015 creator A5082289405 @default.
• W3086152015 date "2020-09-17" @default.
• W3086152015 modified "2023-09-26" @default.
• W3086152015 title "Distilled One-Shot Federated Learning" @default.
• W3086152015 cites W1522301498 @default.
• W3086152015 cites W1533861849 @default.
• W3086152015 cites W1583071929 @default.
• W3086152015 cites W1821462560 @default.
• W3086152015 cites W2090805977 @default.
• W3086152015 cites W2112796928 @default.
• W3086152015 cites W2113459411 @default.
• W3086152015 cites W2127452375 @default.
• W3086152015 cites W2155319834 @default.
• W3086152015 cites W2250539671 @default.
• W3086152015 cites W2283463896 @default.
• W3086152015 cites W2541884796 @default.
• W3086152015 cites W2579186979 @default.
• W3086152015 cites W2604763608 @default.
• W3086152015 cites W2612467436 @default.
• W3086152015 cites W2807006176 @default.
• W3086152015 cites W2914328083 @default.
• W3086152015 cites W2917462349 @default.
• W3086152015 cites W2921434559 @default.
• W3086152015 cites W2943874423 @default.
• W3086152015 cites W2955213239 @default.
• W3086152015 cites W2963902936 @default.
• W3086152015 cites W2970091239 @default.
• W3086152015 cites W2970971581 @default.
• W3086152015 cites W2975043678 @default.
• W3086152015 cites W2979647660 @default.
• W3086152015 cites W2980216952 @default.
• W3086152015 cites W2980910578 @default.
• W3086152015 cites W2981206218 @default.
• W3086152015 cites W2983647115 @default.
• W3086152015 cites W2989289980 @default.
• W3086152015 cites W2995022099 @default.
• W3086152015 cites W2998045710 @default.
• W3086152015 cites W3006555759 @default.
• W3086152015 cites W3008187686 @default.
• W3086152015 cites W3037024761 @default.
• W3086152015 cites W3104920644 @default.
• W3086152015 cites W3137689821 @default.
• W3086152015 cites W3207236107 @default.
• W3086152015 cites W3100125973 @default.
• W3086152015 hasPublicationYear "2020" @default.
• W3086152015 type Work @default.
• W3086152015 sameAs 3086152015 @default.
• W3086152015 citedByCount "13" @default.
• W3086152015 countsByYear W30861520152020 @default.
• W3086152015 countsByYear W30861520152021 @default.
• W3086152015 crossrefType "posted-content" @default.
• W3086152015 hasAuthorship W3086152015A5022526821 @default.
• W3086152015 hasAuthorship W3086152015A5040043382 @default.
• W3086152015 hasAuthorship W3086152015A5061233361 @default.
• W3086152015 hasAuthorship W3086152015A5061695845 @default.
• W3086152015 hasAuthorship W3086152015A5082289405 @default.
• W3086152015 hasConcept C119857082 @default.
• W3086152015 hasConcept C121332964 @default.
• W3086152015 hasConcept C137293760 @default.
• W3086152015 hasConcept C154945302 @default.
• W3086152015 hasConcept C165801399 @default.
• W3086152015 hasConcept C2776788033 @default.
• W3086152015 hasConcept C2992525071 @default.
• W3086152015 hasConcept C31258907 @default.
• W3086152015 hasConcept C41008148 @default.
• W3086152015 hasConcept C62520636 @default.
• W3086152015 hasConcept C66322947 @default.
• W3086152015 hasConceptScore W3086152015C119857082 @default.
• W3086152015 hasConceptScore W3086152015C121332964 @default.
• W3086152015 hasConceptScore W3086152015C137293760 @default.
• W3086152015 hasConceptScore W3086152015C154945302 @default.
• W3086152015 hasConceptScore W3086152015C165801399 @default.
• W3086152015 hasConceptScore W3086152015C2776788033 @default.
• W3086152015 hasConceptScore W3086152015C2992525071 @default.
• W3086152015 hasConceptScore W3086152015C31258907 @default.
• W3086152015 hasConceptScore W3086152015C41008148 @default.
• W3086152015 hasConceptScore W3086152015C62520636 @default.
• W3086152015 hasConceptScore W3086152015C66322947 @default.
• W3086152015 hasLocation W30861520151 @default.
• W3086152015 hasOpenAccess W3086152015 @default.
• W3086152015 hasPrimaryLocation W30861520151 @default.
• W3086152015 hasRelatedWork W1821462560 @default.
• W3086152015 hasRelatedWork W2112796928 @default.
• W3086152015 hasRelatedWork W2194775991 @default.
• W3086152015 hasRelatedWork W2283463896 @default.
• W3086152015 hasRelatedWork W2541884796 @default.
• W3086152015 hasRelatedWork W2777914285 @default.
• W3086152015 hasRelatedWork W2913570153 @default.
• W3086152015 hasRelatedWork W2917462349 @default.
• W3086152015 hasRelatedWork W2918777711 @default.
• W3086152015 hasRelatedWork W2963300197 @default.
• W3086152015 hasRelatedWork W2980216952 @default.
• W3086152015 hasRelatedWork W2995022099 @default.
• W3086152015 hasRelatedWork W3038022836 @default.

• next