FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2946620565> ?p ?o ?g. }

• W2946620565 abstract "In this paper, we introduce the algorithms of Orthogonal Deep Neural Networks (OrthDNNs) to connect with recent interest of spectrally regularized deep learning methods. OrthDNNs are theoretically motivated by generalization analysis of modern DNNs, with the aim to find solution properties of network weights that guarantee better generalization. To this end, we first prove that DNNs are of local isometry on data distributions of practical interest; by using a new covering of the sample space and introducing the local isometry property of DNNs into generalization analysis, we establish a new generalization error bound that is both scale- and range-sensitive to singular value spectrum of each of networks' weight matrices. We prove that the optimal bound w.r.t. the degree of isometry is attained when each weight matrix has a spectrum of equal singular values, among which orthogonal weight matrix or a non-square one with orthonormal rows or columns is the most straightforward choice, suggesting the algorithms of OrthDNNs. We present both algorithms of strict and approximate OrthDNNs, and for the later ones we propose a simple yet effective algorithm called Singular Value Bounding (SVB), which performs as well as strict OrthDNNs, but at a much lower computational cost. We also propose Bounded Batch Normalization (BBN) to make compatible use of batch normalization with OrthDNNs. We conduct extensive comparative studies by using modern architectures on benchmark image classification. Experiments show the efficacy of OrthDNNs." @default.
• W2946620565 created "2019-05-29" @default.
• W2946620565 creator A5001819736 @default.
• W2946620565 creator A5058342436 @default.
• W2946620565 creator A5065250332 @default.
• W2946620565 creator A5065912071 @default.
• W2946620565 creator A5065964089 @default.
• W2946620565 date "2019-05-14" @default.
• W2946620565 modified "2023-09-24" @default.
• W2946620565 title "Orthogonal Deep Neural Networks" @default.
• W2946620565 cites W104184427 @default.
• W2946620565 cites W1533861849 @default.
• W2946620565 cites W1804110266 @default.
• W2946620565 cites W1836465849 @default.
• W2946620565 cites W1892947258 @default.
• W2946620565 cites W1904365287 @default.
• W2946620565 cites W1944672 @default.
• W2946620565 cites W2101926813 @default.
• W2946620565 cites W2112796928 @default.
• W2946620565 cites W2129131372 @default.
• W2946620565 cites W2139338362 @default.
• W2946620565 cites W2152722485 @default.
• W2946620565 cites W2153072115 @default.
• W2946620565 cites W2156387975 @default.
• W2946620565 cites W2161388792 @default.
• W2946620565 cites W2166116275 @default.
• W2946620565 cites W2175402905 @default.
• W2946620565 cites W2181330269 @default.
• W2946620565 cites W2194775991 @default.
• W2946620565 cites W2294567968 @default.
• W2946620565 cites W2302255633 @default.
• W2946620565 cites W2401231614 @default.
• W2946620565 cites W2473476732 @default.
• W2946620565 cites W2529714286 @default.
• W2946620565 cites W2546257475 @default.
• W2946620565 cites W2549139847 @default.
• W2946620565 cites W2552391848 @default.
• W2946620565 cites W2556882396 @default.
• W2946620565 cites W2593110912 @default.
• W2946620565 cites W2597984439 @default.
• W2946620565 cites W2604451472 @default.
• W2946620565 cites W2605372163 @default.
• W2946620565 cites W2662969263 @default.
• W2946620565 cites W2766196653 @default.
• W2946620565 cites W2914484425 @default.
• W2946620565 cites W2949292735 @default.
• W2946620565 cites W2951605425 @default.
• W2946620565 cites W2953106684 @default.
• W2946620565 cites W2962724006 @default.
• W2946620565 cites W2962810483 @default.
• W2946620565 cites W2962835968 @default.
• W2946620565 cites W2963060032 @default.
• W2946620565 cites W2963103976 @default.
• W2946620565 cites W2963424732 @default.
• W2946620565 cites W2963446085 @default.
• W2946620565 cites W2963446712 @default.
• W2946620565 cites W2963504252 @default.
• W2946620565 cites W2963570896 @default.
• W2946620565 cites W2963586744 @default.
• W2946620565 cites W2963606038 @default.
• W2946620565 cites W2963794891 @default.
• W2946620565 cites W2963959597 @default.
• W2946620565 cites W2964121744 @default.
• W2946620565 cites W2964294232 @default.
• W2946620565 cites W3102317997 @default.
• W2946620565 cites W3118608800 @default.
• W2946620565 cites W3137695714 @default.
• W2946620565 cites W3146803896 @default.
• W2946620565 cites W377884864 @default.
• W2946620565 doi "https://doi.org/10.48550/arxiv.1905.05929" @default.
• W2946620565 hasPublicationYear "2019" @default.
• W2946620565 type Work @default.
• W2946620565 sameAs 2946620565 @default.
• W2946620565 citedByCount "6" @default.
• W2946620565 countsByYear W29466205652019 @default.
• W2946620565 countsByYear W29466205652020 @default.
• W2946620565 crossrefType "posted-content" @default.
• W2946620565 hasAuthorship W2946620565A5001819736 @default.
• W2946620565 hasAuthorship W2946620565A5058342436 @default.
• W2946620565 hasAuthorship W2946620565A5065250332 @default.
• W2946620565 hasAuthorship W2946620565A5065912071 @default.
• W2946620565 hasAuthorship W2946620565A5065964089 @default.
• W2946620565 hasBestOaLocation W29466205651 @default.
• W2946620565 hasConcept C106487976 @default.
• W2946620565 hasConcept C109282560 @default.
• W2946620565 hasConcept C11413529 @default.
• W2946620565 hasConcept C121332964 @default.
• W2946620565 hasConcept C124851039 @default.
• W2946620565 hasConcept C134306372 @default.
• W2946620565 hasConcept C136886441 @default.
• W2946620565 hasConcept C144024400 @default.
• W2946620565 hasConcept C154945302 @default.
• W2946620565 hasConcept C158693339 @default.
• W2946620565 hasConcept C159985019 @default.
• W2946620565 hasConcept C177148314 @default.
• W2946620565 hasConcept C17902559 @default.
• W2946620565 hasConcept C19165224 @default.
• W2946620565 hasConcept C192562407 @default.
• W2946620565 hasConcept C33923547 @default.
• W2946620565 hasConcept C34388435 @default.

• next