Matches in SemOpenAlex for { <https://semopenalex.org/work/W2901161862> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W2901161862 abstract "The world needs diverse and unbiased data to train deep learning models. Currently data comes from a variety of sources that are unmoderated to a large extent. The outcomes of training neural networks with unverified data yields biased models with various strains of homophobia, sexism and racism. Another trend observed in the world of deep learning is the rise of distributed training. Although cloud companies provide high performance compute for training models in the form of GPU's connected with a low latency network, using these services comes at a high cost. We propose Hydra, a system that seeks to solve both of these problems in a novel manner by proposing a decentralized distributed framework which utilizes the substantial amount of idle compute of everyday electronic devices like smartphones and desktop computers for training and data collection purposes. Hydra couples a specialized distributed training framework on a network of these low powered devices with a reward scheme that incentivizes users to provide high quality data to unleash the compute capability on this training framework. Such a system has the ability to capture data from a wide variety of diverse sources which has been an issue in the current scenario of deep learning. Hydra brings in several new innovations in training on low powered devices including a fault tolerant version of the All Reduce algorithm. Furthermore we introduce a reinforcement learning policy to decide the size of training jobs on different machines on a heterogeneous cluster of devices with varying network latencies for Synchronous SGD. The novel thing about such a network is the ability of each machine to shut down and resume training capabilities at any point of time without restarting the overall training. To enable such an asynchronous behaviour we propose a communication framework inspired by the Bittorrent protocol and the Kademlia DHT." @default.
- W2901161862 created "2018-11-29" @default.
- W2901161862 creator A5072233594 @default.
- W2901161862 creator A5075458587 @default.
- W2901161862 date "2018-11-24" @default.
- W2901161862 modified "2023-09-27" @default.
- W2901161862 title "Hydra: A Peer to Peer Distributed Training & Data Collection Framework." @default.
- W2901161862 cites W2108598243 @default.
- W2901161862 cites W2119245106 @default.
- W2901161862 cites W2119717200 @default.
- W2901161862 cites W2156580773 @default.
- W2901161862 cites W2182688186 @default.
- W2901161862 cites W2336650964 @default.
- W2901161862 cites W2523246573 @default.
- W2901161862 cites W2553303224 @default.
- W2901161862 cites W2622263826 @default.
- W2901161862 cites W2763421725 @default.
- W2901161862 cites W2766164908 @default.
- W2901161862 cites W2785717964 @default.
- W2901161862 cites W2787998955 @default.
- W2901161862 cites W2884711234 @default.
- W2901161862 cites W2900319533 @default.
- W2901161862 cites W2949952845 @default.
- W2901161862 cites W2952926545 @default.
- W2901161862 cites W2963803379 @default.
- W2901161862 cites W3124258878 @default.
- W2901161862 hasPublicationYear "2018" @default.
- W2901161862 type Work @default.
- W2901161862 sameAs 2901161862 @default.
- W2901161862 citedByCount "2" @default.
- W2901161862 countsByYear W29011618622020 @default.
- W2901161862 crossrefType "posted-content" @default.
- W2901161862 hasAuthorship W2901161862A5072233594 @default.
- W2901161862 hasAuthorship W2901161862A5075458587 @default.
- W2901161862 hasConcept C105795698 @default.
- W2901161862 hasConcept C108583219 @default.
- W2901161862 hasConcept C119857082 @default.
- W2901161862 hasConcept C120314980 @default.
- W2901161862 hasConcept C121332964 @default.
- W2901161862 hasConcept C133462117 @default.
- W2901161862 hasConcept C136197465 @default.
- W2901161862 hasConcept C153294291 @default.
- W2901161862 hasConcept C153740404 @default.
- W2901161862 hasConcept C154945302 @default.
- W2901161862 hasConcept C2522767166 @default.
- W2901161862 hasConcept C2777211547 @default.
- W2901161862 hasConcept C31258907 @default.
- W2901161862 hasConcept C33923547 @default.
- W2901161862 hasConcept C41008148 @default.
- W2901161862 hasConcept C50644808 @default.
- W2901161862 hasConcept C63540848 @default.
- W2901161862 hasConcept C97541855 @default.
- W2901161862 hasConceptScore W2901161862C105795698 @default.
- W2901161862 hasConceptScore W2901161862C108583219 @default.
- W2901161862 hasConceptScore W2901161862C119857082 @default.
- W2901161862 hasConceptScore W2901161862C120314980 @default.
- W2901161862 hasConceptScore W2901161862C121332964 @default.
- W2901161862 hasConceptScore W2901161862C133462117 @default.
- W2901161862 hasConceptScore W2901161862C136197465 @default.
- W2901161862 hasConceptScore W2901161862C153294291 @default.
- W2901161862 hasConceptScore W2901161862C153740404 @default.
- W2901161862 hasConceptScore W2901161862C154945302 @default.
- W2901161862 hasConceptScore W2901161862C2522767166 @default.
- W2901161862 hasConceptScore W2901161862C2777211547 @default.
- W2901161862 hasConceptScore W2901161862C31258907 @default.
- W2901161862 hasConceptScore W2901161862C33923547 @default.
- W2901161862 hasConceptScore W2901161862C41008148 @default.
- W2901161862 hasConceptScore W2901161862C50644808 @default.
- W2901161862 hasConceptScore W2901161862C63540848 @default.
- W2901161862 hasConceptScore W2901161862C97541855 @default.
- W2901161862 hasLocation W29011618621 @default.
- W2901161862 hasOpenAccess W2901161862 @default.
- W2901161862 hasPrimaryLocation W29011618621 @default.
- W2901161862 hasRelatedWork W2901256196 @default.
- W2901161862 hasRelatedWork W2911714797 @default.
- W2901161862 hasRelatedWork W2918828872 @default.
- W2901161862 hasRelatedWork W2945433136 @default.
- W2901161862 hasRelatedWork W2975118496 @default.
- W2901161862 hasRelatedWork W2975942746 @default.
- W2901161862 hasRelatedWork W2982592664 @default.
- W2901161862 hasRelatedWork W3011123158 @default.
- W2901161862 hasRelatedWork W3016216736 @default.
- W2901161862 hasRelatedWork W3102606411 @default.
- W2901161862 hasRelatedWork W3116992856 @default.
- W2901161862 hasRelatedWork W3133480365 @default.
- W2901161862 hasRelatedWork W3135099698 @default.
- W2901161862 hasRelatedWork W3137220449 @default.
- W2901161862 hasRelatedWork W3137518503 @default.
- W2901161862 hasRelatedWork W3164522852 @default.
- W2901161862 hasRelatedWork W3167470315 @default.
- W2901161862 hasRelatedWork W3200219902 @default.
- W2901161862 hasRelatedWork W3204105474 @default.
- W2901161862 hasRelatedWork W3209080999 @default.
- W2901161862 isParatext "false" @default.
- W2901161862 isRetracted "false" @default.
- W2901161862 magId "2901161862" @default.
- W2901161862 workType "article" @default.