FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3201712558> ?p ?o ?g. }

• W3201712558 abstract "Despite their success deep neural networks (DNNs) are still largely considered as black boxes. The main issue is that the linear and non-linear operations are entangled in every layer, making it hard to interpret the hidden layer outputs. In this paper, we look at DNNs with rectified linear units (ReLUs), and focus on the gating property (`on/off' states) of the ReLUs. We extend the recently developed dual view in which the computation is broken path-wise to show that learning in the gates is more crucial, and learning the weights given the gates is characterised analytically via the so called neural path kernel (NPK) which depends on inputs and gates. In this paper, we present novel results to show that convolution with global pooling and skip connection provide respectively rotational invariance and ensemble structure to the NPK. To address `black box'-ness, we propose a novel interpretable counterpart of DNNs with ReLUs namely deep linearly gated networks (DLGN): the pre-activations to the gates are generated by a deep linear network, and the gates are then applied as external masks to learn the weights in a different network. The DLGN is not an alternative architecture per se, but a disentanglement and an interpretable re-arrangement of the computations in a DNN with ReLUs. The DLGN disentangles the computations into two `mathematically' interpretable linearities (i) the `primal' linearity between the input and the pre-activations in the gating network and (ii) the `dual' linearity in the path space in the weights network characterised by the NPK. We compare the performance of DNN, DGN and DLGN on CIFAR-10 and CIFAR-100 to show that, the DLGN recovers more than $83.5%$ of the performance of state-of-the-art DNNs. This brings us to an interesting question: `Is DLGN a universal spectral approximator?'" @default.
• W3201712558 created "2021-10-11" @default.
• W3201712558 creator A5044247596 @default.
• W3201712558 creator A5067909461 @default.
• W3201712558 date "2021-10-06" @default.
• W3201712558 modified "2023-09-27" @default.
• W3201712558 title "Disentangling deep neural networks with rectified linear units using duality." @default.
• W3201712558 cites W1522301498 @default.
• W3201712558 cites W2809090039 @default.
• W3201712558 cites W2910655610 @default.
• W3201712558 cites W2923179079 @default.
• W3201712558 cites W2962766031 @default.
• W3201712558 cites W2963410064 @default.
• W3201712558 cites W2964052793 @default.
• W3201712558 cites W2970332347 @default.
• W3201712558 cites W2971043187 @default.
• W3201712558 cites W2987473824 @default.
• W3201712558 cites W3099571327 @default.
• W3201712558 cites W3099842580 @default.
• W3201712558 cites W3101069636 @default.
• W3201712558 cites W3102429474 @default.
• W3201712558 cites W3104810215 @default.
• W3201712558 cites W3105415784 @default.
• W3201712558 cites W3137695714 @default.
• W3201712558 cites W2803403308 @default.
• W3201712558 cites W3034194698 @default.
• W3201712558 hasPublicationYear "2021" @default.
• W3201712558 type Work @default.
• W3201712558 sameAs 3201712558 @default.
• W3201712558 citedByCount "0" @default.
• W3201712558 crossrefType "posted-content" @default.
• W3201712558 hasAuthorship W3201712558A5044247596 @default.
• W3201712558 hasAuthorship W3201712558A5067909461 @default.
• W3201712558 hasConcept C108583219 @default.
• W3201712558 hasConcept C11413529 @default.
• W3201712558 hasConcept C118615104 @default.
• W3201712558 hasConcept C120665830 @default.
• W3201712558 hasConcept C121332964 @default.
• W3201712558 hasConcept C134306372 @default.
• W3201712558 hasConcept C154945302 @default.
• W3201712558 hasConcept C177148314 @default.
• W3201712558 hasConcept C192209626 @default.
• W3201712558 hasConcept C194544171 @default.
• W3201712558 hasConcept C199360897 @default.
• W3201712558 hasConcept C2777735758 @default.
• W3201712558 hasConcept C2984842247 @default.
• W3201712558 hasConcept C33923547 @default.
• W3201712558 hasConcept C41008148 @default.
• W3201712558 hasConcept C42407357 @default.
• W3201712558 hasConcept C45347329 @default.
• W3201712558 hasConcept C45374587 @default.
• W3201712558 hasConcept C50644808 @default.
• W3201712558 hasConcept C62520636 @default.
• W3201712558 hasConcept C70437156 @default.
• W3201712558 hasConcept C74193536 @default.
• W3201712558 hasConcept C77170095 @default.
• W3201712558 hasConcept C81363708 @default.
• W3201712558 hasConcept C86803240 @default.
• W3201712558 hasConceptScore W3201712558C108583219 @default.
• W3201712558 hasConceptScore W3201712558C11413529 @default.
• W3201712558 hasConceptScore W3201712558C118615104 @default.
• W3201712558 hasConceptScore W3201712558C120665830 @default.
• W3201712558 hasConceptScore W3201712558C121332964 @default.
• W3201712558 hasConceptScore W3201712558C134306372 @default.
• W3201712558 hasConceptScore W3201712558C154945302 @default.
• W3201712558 hasConceptScore W3201712558C177148314 @default.
• W3201712558 hasConceptScore W3201712558C192209626 @default.
• W3201712558 hasConceptScore W3201712558C194544171 @default.
• W3201712558 hasConceptScore W3201712558C199360897 @default.
• W3201712558 hasConceptScore W3201712558C2777735758 @default.
• W3201712558 hasConceptScore W3201712558C2984842247 @default.
• W3201712558 hasConceptScore W3201712558C33923547 @default.
• W3201712558 hasConceptScore W3201712558C41008148 @default.
• W3201712558 hasConceptScore W3201712558C42407357 @default.
• W3201712558 hasConceptScore W3201712558C45347329 @default.
• W3201712558 hasConceptScore W3201712558C45374587 @default.
• W3201712558 hasConceptScore W3201712558C50644808 @default.
• W3201712558 hasConceptScore W3201712558C62520636 @default.
• W3201712558 hasConceptScore W3201712558C70437156 @default.
• W3201712558 hasConceptScore W3201712558C74193536 @default.
• W3201712558 hasConceptScore W3201712558C77170095 @default.
• W3201712558 hasConceptScore W3201712558C81363708 @default.
• W3201712558 hasConceptScore W3201712558C86803240 @default.
• W3201712558 hasLocation W32017125581 @default.
• W3201712558 hasOpenAccess W3201712558 @default.
• W3201712558 hasPrimaryLocation W32017125581 @default.
• W3201712558 hasRelatedWork W2005971284 @default.
• W3201712558 hasRelatedWork W2161388792 @default.
• W3201712558 hasRelatedWork W2547432194 @default.
• W3201712558 hasRelatedWork W2593301408 @default.
• W3201712558 hasRelatedWork W2618540739 @default.
• W3201712558 hasRelatedWork W2735918603 @default.
• W3201712558 hasRelatedWork W2766956386 @default.
• W3201712558 hasRelatedWork W2790252067 @default.
• W3201712558 hasRelatedWork W2892830725 @default.
• W3201712558 hasRelatedWork W2905333383 @default.
• W3201712558 hasRelatedWork W2978030971 @default.
• W3201712558 hasRelatedWork W2978469940 @default.
• W3201712558 hasRelatedWork W3004565269 @default.
• W3201712558 hasRelatedWork W3005559199 @default.

• next