FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2948161566> ?p ?o ?g. }

• W2948161566 abstract "Deep learning relies on good initialization schemes and hyperparameter choices prior to training a neural network. Random weight initializations induce random network ensembles, which give rise to the trainability, training speed, and sometimes also generalization ability of an instance. In addition, such ensembles provide theoretical insights into the space of candidate models of which one is selected during training. The results obtained so far rely on mean field approximations that assume infinite layer width and that study average squared signals. We derive the joint signal output distribution exactly, without mean field assumptions, for fully-connected networks with Gaussian weights and biases, and analyze deviations from the mean field results. For rectified linear units, we further discuss limitations of the standard initialization scheme, such as its lack of dynamical isometry, and propose a simple alternative that overcomes these by initial parameter sharing." @default.
• W2948161566 created "2019-06-14" @default.
• W2948161566 creator A5076324337 @default.
• W2948161566 creator A5076506988 @default.
• W2948161566 date "2018-06-17" @default.
• W2948161566 modified "2023-09-23" @default.
• W2948161566 title "Initialization of ReLUs for Dynamical Isometry" @default.
• W2948161566 cites W1533861849 @default.
• W2948161566 cites W1677182931 @default.
• W2948161566 cites W1836465849 @default.
• W2948161566 cites W2112796928 @default.
• W2948161566 cites W2149933564 @default.
• W2948161566 cites W2167608136 @default.
• W2948161566 cites W2591954064 @default.
• W2948161566 cites W2891962546 @default.
• W2948161566 cites W2962685937 @default.
• W2948161566 cites W2962804662 @default.
• W2948161566 cites W2962930287 @default.
• W2948161566 cites W2963037478 @default.
• W2948161566 cites W2963063862 @default.
• W2948161566 cites W2963148870 @default.
• W2948161566 cites W2963504252 @default.
• W2948161566 cites W2963570896 @default.
• W2948161566 cites W2963966020 @default.
• W2948161566 cites W2963982496 @default.
• W2948161566 cites W2964003773 @default.
• W2948161566 cites W2964050767 @default.
• W2948161566 cites W2964052793 @default.
• W2948161566 cites W2964065616 @default.
• W2948161566 cites W2964088238 @default.
• W2948161566 cites W2964161337 @default.
• W2948161566 cites W4919037 @default.
• W2948161566 hasPublicationYear "2018" @default.
• W2948161566 type Work @default.
• W2948161566 sameAs 2948161566 @default.
• W2948161566 citedByCount "0" @default.
• W2948161566 crossrefType "posted-content" @default.
• W2948161566 hasAuthorship W2948161566A5076324337 @default.
• W2948161566 hasAuthorship W2948161566A5076506988 @default.
• W2948161566 hasConcept C11413529 @default.
• W2948161566 hasConcept C114466953 @default.
• W2948161566 hasConcept C121332964 @default.
• W2948161566 hasConcept C134306372 @default.
• W2948161566 hasConcept C154945302 @default.
• W2948161566 hasConcept C162324750 @default.
• W2948161566 hasConcept C163716315 @default.
• W2948161566 hasConcept C176217482 @default.
• W2948161566 hasConcept C177148314 @default.
• W2948161566 hasConcept C199360897 @default.
• W2948161566 hasConcept C202444582 @default.
• W2948161566 hasConcept C21547014 @default.
• W2948161566 hasConcept C28826006 @default.
• W2948161566 hasConcept C33923547 @default.
• W2948161566 hasConcept C41008148 @default.
• W2948161566 hasConcept C50644808 @default.
• W2948161566 hasConcept C62520636 @default.
• W2948161566 hasConcept C82457910 @default.
• W2948161566 hasConcept C8642999 @default.
• W2948161566 hasConcept C9652623 @default.
• W2948161566 hasConceptScore W2948161566C11413529 @default.
• W2948161566 hasConceptScore W2948161566C114466953 @default.
• W2948161566 hasConceptScore W2948161566C121332964 @default.
• W2948161566 hasConceptScore W2948161566C134306372 @default.
• W2948161566 hasConceptScore W2948161566C154945302 @default.
• W2948161566 hasConceptScore W2948161566C162324750 @default.
• W2948161566 hasConceptScore W2948161566C163716315 @default.
• W2948161566 hasConceptScore W2948161566C176217482 @default.
• W2948161566 hasConceptScore W2948161566C177148314 @default.
• W2948161566 hasConceptScore W2948161566C199360897 @default.
• W2948161566 hasConceptScore W2948161566C202444582 @default.
• W2948161566 hasConceptScore W2948161566C21547014 @default.
• W2948161566 hasConceptScore W2948161566C28826006 @default.
• W2948161566 hasConceptScore W2948161566C33923547 @default.
• W2948161566 hasConceptScore W2948161566C41008148 @default.
• W2948161566 hasConceptScore W2948161566C50644808 @default.
• W2948161566 hasConceptScore W2948161566C62520636 @default.
• W2948161566 hasConceptScore W2948161566C82457910 @default.
• W2948161566 hasConceptScore W2948161566C8642999 @default.
• W2948161566 hasConceptScore W2948161566C9652623 @default.
• W2948161566 hasOpenAccess W2948161566 @default.
• W2948161566 hasRelatedWork W1531921002 @default.
• W2948161566 hasRelatedWork W1756018955 @default.
• W2948161566 hasRelatedWork W2148474239 @default.
• W2948161566 hasRelatedWork W2472226770 @default.
• W2948161566 hasRelatedWork W2608781926 @default.
• W2948161566 hasRelatedWork W2766635461 @default.
• W2948161566 hasRelatedWork W2903666595 @default.
• W2948161566 hasRelatedWork W2918745211 @default.
• W2948161566 hasRelatedWork W2941256613 @default.
• W2948161566 hasRelatedWork W2952979223 @default.
• W2948161566 hasRelatedWork W2963161247 @default.
• W2948161566 hasRelatedWork W2970375336 @default.
• W2948161566 hasRelatedWork W2998099803 @default.
• W2948161566 hasRelatedWork W3034711337 @default.
• W2948161566 hasRelatedWork W3119796929 @default.
• W2948161566 hasRelatedWork W3126823575 @default.
• W2948161566 hasRelatedWork W3157298807 @default.
• W2948161566 hasRelatedWork W3181674477 @default.
• W2948161566 hasRelatedWork W3202410655 @default.
• W2948161566 hasRelatedWork W3211851805 @default.

• next