FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2178314882> ?p ?o ?g. }

• W2178314882 abstract "Recent success in training deep neural networks have prompted active investigation into the features learned on their intermediate layers. Such research is difficult because it requires making sense of non-linear computations performed by millions of parameters, but valuable because it increases our ability to understand current models and create improved versions of them. In this paper we investigate the extent to which neural networks exhibit what we call convergent learning, which is when the representations learned by multiple nets converge to a set of features which are either individually similar between networks or where subsets of features span similar low-dimensional spaces. We propose a specific method of probing representations: training multiple networks and then comparing and contrasting their individual, learned representations at the level of neurons or groups of neurons. We begin research into this question using three techniques to approximately align different neural networks on a feature level: a bipartite matching approach that makes one-to-one assignments between neurons, a sparse prediction approach that finds one-to-many mappings, and a spectral clustering approach that finds many-to-many mappings. This initial investigation reveals a few previously unknown properties of neural networks, and we argue that future research into the question of convergent learning will yield many more. The insights described here include (1) that some features are learned reliably in multiple networks, yet other features are not consistently learned; (2) that units learn to span low-dimensional subspaces and, while these subspaces are common to multiple networks, the specific basis vectors learned are not; (3) that the representation codes show evidence of being a mix between a local code and slightly, but not fully, distributed codes across multiple units." @default.
• W2178314882 created "2016-06-24" @default.
• W2178314882 creator A5006797018 @default.
• W2178314882 creator A5025894735 @default.
• W2178314882 creator A5043211393 @default.
• W2178314882 creator A5049659753 @default.
• W2178314882 creator A5078018786 @default.
• W2178314882 date "2015-11-24" @default.
• W2178314882 modified "2023-10-02" @default.
• W2178314882 title "Convergent Learning: Do different neural networks learn the same representations?" @default.
• W2178314882 cites W1493372406 @default.
• W2178314882 cites W1496253510 @default.
• W2178314882 cites W1522294902 @default.
• W2178314882 cites W1673923490 @default.
• W2178314882 cites W1825675169 @default.
• W2178314882 cites W1849277567 @default.
• W2178314882 cites W1912570122 @default.
• W2178314882 cites W1932198206 @default.
• W2178314882 cites W2101234009 @default.
• W2178314882 cites W2108598243 @default.
• W2178314882 cites W2125690626 @default.
• W2178314882 cites W2135046866 @default.
• W2178314882 cites W2143915663 @default.
• W2178314882 cites W2146989110 @default.
• W2178314882 cites W2157529519 @default.
• W2178314882 cites W2169393322 @default.
• W2178314882 cites W2949667497 @default.
• W2178314882 cites W2949987032 @default.
• W2178314882 cites W2950094539 @default.
• W2178314882 cites W2963025229 @default.
• W2178314882 hasPublicationYear "2015" @default.
• W2178314882 type Work @default.
• W2178314882 sameAs 2178314882 @default.
• W2178314882 citedByCount "30" @default.
• W2178314882 countsByYear W21783148822016 @default.
• W2178314882 countsByYear W21783148822017 @default.
• W2178314882 countsByYear W21783148822018 @default.
• W2178314882 countsByYear W21783148822019 @default.
• W2178314882 countsByYear W21783148822020 @default.
• W2178314882 countsByYear W21783148822021 @default.
• W2178314882 crossrefType "posted-content" @default.
• W2178314882 hasAuthorship W2178314882A5006797018 @default.
• W2178314882 hasAuthorship W2178314882A5025894735 @default.
• W2178314882 hasAuthorship W2178314882A5043211393 @default.
• W2178314882 hasAuthorship W2178314882A5049659753 @default.
• W2178314882 hasAuthorship W2178314882A5078018786 @default.
• W2178314882 hasConcept C105795698 @default.
• W2178314882 hasConcept C108583219 @default.
• W2178314882 hasConcept C119857082 @default.
• W2178314882 hasConcept C12362212 @default.
• W2178314882 hasConcept C132525143 @default.
• W2178314882 hasConcept C138885662 @default.
• W2178314882 hasConcept C154945302 @default.
• W2178314882 hasConcept C165064840 @default.
• W2178314882 hasConcept C177264268 @default.
• W2178314882 hasConcept C197657726 @default.
• W2178314882 hasConcept C199360897 @default.
• W2178314882 hasConcept C2524010 @default.
• W2178314882 hasConcept C2776401178 @default.
• W2178314882 hasConcept C33923547 @default.
• W2178314882 hasConcept C41008148 @default.
• W2178314882 hasConcept C41895202 @default.
• W2178314882 hasConcept C50644808 @default.
• W2178314882 hasConcept C80444323 @default.
• W2178314882 hasConceptScore W2178314882C105795698 @default.
• W2178314882 hasConceptScore W2178314882C108583219 @default.
• W2178314882 hasConceptScore W2178314882C119857082 @default.
• W2178314882 hasConceptScore W2178314882C12362212 @default.
• W2178314882 hasConceptScore W2178314882C132525143 @default.
• W2178314882 hasConceptScore W2178314882C138885662 @default.
• W2178314882 hasConceptScore W2178314882C154945302 @default.
• W2178314882 hasConceptScore W2178314882C165064840 @default.
• W2178314882 hasConceptScore W2178314882C177264268 @default.
• W2178314882 hasConceptScore W2178314882C197657726 @default.
• W2178314882 hasConceptScore W2178314882C199360897 @default.
• W2178314882 hasConceptScore W2178314882C2524010 @default.
• W2178314882 hasConceptScore W2178314882C2776401178 @default.
• W2178314882 hasConceptScore W2178314882C33923547 @default.
• W2178314882 hasConceptScore W2178314882C41008148 @default.
• W2178314882 hasConceptScore W2178314882C41895202 @default.
• W2178314882 hasConceptScore W2178314882C50644808 @default.
• W2178314882 hasConceptScore W2178314882C80444323 @default.
• W2178314882 hasLocation W21783148821 @default.
• W2178314882 hasOpenAccess W2178314882 @default.
• W2178314882 hasPrimaryLocation W21783148821 @default.
• W2178314882 hasRelatedWork W1686810756 @default.
• W2178314882 hasRelatedWork W1825675169 @default.
• W2178314882 hasRelatedWork W1849277567 @default.
• W2178314882 hasRelatedWork W1951216520 @default.
• W2178314882 hasRelatedWork W2009284521 @default.
• W2178314882 hasRelatedWork W2095705004 @default.
• W2178314882 hasRelatedWork W2097117768 @default.
• W2178314882 hasRelatedWork W2117539524 @default.
• W2178314882 hasRelatedWork W2163605009 @default.
• W2178314882 hasRelatedWork W2194775991 @default.
• W2178314882 hasRelatedWork W2418098761 @default.
• W2178314882 hasRelatedWork W2767204723 @default.
• W2178314882 hasRelatedWork W2898482221 @default.
• W2178314882 hasRelatedWork W2919115771 @default.
• W2178314882 hasRelatedWork W2949117887 @default.

• next