FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3128694157> ?p ?o ?g. }

• W3128694157 abstract "Conventional neural architectures for sequential data present important limitations. Recurrent networks suffer from exploding and vanishing gradients, small effective memory horizons, and must be trained sequentially. Convolutional networks are unable to handle sequences of unknown size and their memory horizon must be defined a priori. In this work, we show that all these problems can be solved by formulating convolutional kernels in CNNs as continuous functions. The resulting Continuous Kernel Convolution (CKConv) allows us to model arbitrarily long sequences in a parallel manner, within a single operation, and without relying on any form of recurrence. We show that Continuous Kernel Convolutional Networks (CKCNNs) obtain state-of-the-art results in multiple datasets, e.g., permuted MNIST, and, thanks to their continuous nature, are able to handle non-uniformly sampled datasets and irregularly-sampled data natively. CKCNNs match or perform better than neural ODEs designed for these purposes in a much faster and simpler manner." @default.
• W3128694157 created "2021-02-15" @default.
• W3128694157 creator A5000647841 @default.
• W3128694157 creator A5004385039 @default.
• W3128694157 creator A5044217809 @default.
• W3128694157 creator A5058054844 @default.
• W3128694157 creator A5084474296 @default.
• W3128694157 date "2021-02-04" @default.
• W3128694157 modified "2023-09-26" @default.
• W3128694157 title "CKConv: Continuous Kernel Convolution For Sequential Data" @default.
• W3128694157 cites W1522301498 @default.
• W3128694157 cites W1665214252 @default.
• W3128694157 cites W1800356822 @default.
• W3128694157 cites W1815076433 @default.
• W3128694157 cites W1889624880 @default.
• W3128694157 cites W1921523184 @default.
• W3128694157 cites W1924770834 @default.
• W3128694157 cites W2064675550 @default.
• W3128694157 cites W2107878631 @default.
• W3128694157 cites W2112796928 @default.
• W3128694157 cites W2143612262 @default.
• W3128694157 cites W2150355110 @default.
• W3128694157 cites W2161388792 @default.
• W3128694157 cites W2504108613 @default.
• W3128694157 cites W2529337537 @default.
• W3128694157 cites W2606202972 @default.
• W3128694157 cites W2766736793 @default.
• W3128694157 cites W2767286248 @default.
• W3128694157 cites W2788775653 @default.
• W3128694157 cites W2792764867 @default.
• W3128694157 cites W2793273050 @default.
• W3128694157 cites W2797583228 @default.
• W3128694157 cites W2884447957 @default.
• W3128694157 cites W2896528354 @default.
• W3128694157 cites W2898647012 @default.
• W3128694157 cites W2911109671 @default.
• W3128694157 cites W2914344989 @default.
• W3128694157 cites W2949382160 @default.
• W3128694157 cites W2949718784 @default.
• W3128694157 cites W2949807021 @default.
• W3128694157 cites W2963042606 @default.
• W3128694157 cites W2963231572 @default.
• W3128694157 cites W2963241221 @default.
• W3128694157 cites W2963627347 @default.
• W3128694157 cites W2963685250 @default.
• W3128694157 cites W2963711743 @default.
• W3128694157 cites W2963926543 @default.
• W3128694157 cites W2963950090 @default.
• W3128694157 cites W2963970792 @default.
• W3128694157 cites W2964010366 @default.
• W3128694157 cites W2964046515 @default.
• W3128694157 cites W2964347220 @default.
• W3128694157 cites W2968949658 @default.
• W3128694157 cites W2971278153 @default.
• W3128694157 cites W2971306341 @default.
• W3128694157 cites W3008105217 @default.
• W3128694157 cites W3008803199 @default.
• W3128694157 cites W3011537551 @default.
• W3128694157 cites W3012494314 @default.
• W3128694157 cites W3034504596 @default.
• W3128694157 cites W3034576161 @default.
• W3128694157 cites W3035332875 @default.
• W3128694157 cites W3036737467 @default.
• W3128694157 cites W3036843665 @default.
• W3128694157 cites W3089607359 @default.
• W3128694157 cites W3096986660 @default.
• W3128694157 cites W3101267796 @default.
• W3128694157 cites W3103313582 @default.
• W3128694157 cites W3114474797 @default.
• W3128694157 cites W3115295152 @default.
• W3128694157 cites W3118608800 @default.
• W3128694157 cites W3126517808 @default.
• W3128694157 cites W194249466 @default.
• W3128694157 cites W3099512283 @default.
• W3128694157 hasPublicationYear "2021" @default.
• W3128694157 type Work @default.
• W3128694157 sameAs 3128694157 @default.
• W3128694157 citedByCount "8" @default.
• W3128694157 countsByYear W31286941572021 @default.
• W3128694157 crossrefType "posted-content" @default.
• W3128694157 hasAuthorship W3128694157A5000647841 @default.
• W3128694157 hasAuthorship W3128694157A5004385039 @default.
• W3128694157 hasAuthorship W3128694157A5044217809 @default.
• W3128694157 hasAuthorship W3128694157A5058054844 @default.
• W3128694157 hasAuthorship W3128694157A5084474296 @default.
• W3128694157 hasConcept C108583219 @default.
• W3128694157 hasConcept C111472728 @default.
• W3128694157 hasConcept C11413529 @default.
• W3128694157 hasConcept C118615104 @default.
• W3128694157 hasConcept C138885662 @default.
• W3128694157 hasConcept C153180895 @default.
• W3128694157 hasConcept C154945302 @default.
• W3128694157 hasConcept C190502265 @default.
• W3128694157 hasConcept C33923547 @default.
• W3128694157 hasConcept C41008148 @default.
• W3128694157 hasConcept C45347329 @default.
• W3128694157 hasConcept C50644808 @default.
• W3128694157 hasConcept C74193536 @default.
• W3128694157 hasConcept C75553542 @default.
• W3128694157 hasConcept C81363708 @default.

• next