Matches in SemOpenAlex for { <https://semopenalex.org/work/W3037039100> ?p ?o ?g. }
Showing items 1 to 84 of
84
with 100 items per page.
- W3037039100 endingPage "515" @default.
- W3037039100 startingPage "504" @default.
- W3037039100 abstract "Deep neural networks generalize well on unseen data though the number of parameters often far exceeds the number of training examples. Recently proposed complexity measures have provided insights to understanding the generalizability in neural networks from perspectives of PAC-Bayes, robustness, overparametrization, compression and so on. In this work, we advance the understanding of the relations between the network's architecture and its generalizability from the compression perspective. Using tensor analysis, we propose a series of intuitive, data-dependent and easily-measurable properties that tightly characterize the compressibility and generalizability of neural networks; thus, in practice, our generalization bound outperforms the previous compression-based ones, especially for neural networks using tensors as their weight kernels (e.g. CNNs). Moreover, these intuitive measurements provide further insights into designing neural network architectures with properties favorable for better/guaranteed generalizability. Our experimental results demonstrate that through the proposed measurable properties, our generalization error bound matches the trend of the test error well. Our theoretical analysis further provides justifications for the empirical success and limitations of some widely-used tensor-based compression approaches. We also discover the improvements to the compressibility and robustness of current neural networks when incorporating tensor operations via our proposed layer-wise structure." @default.
- W3037039100 created "2020-07-02" @default.
- W3037039100 creator A5008267969 @default.
- W3037039100 creator A5010098535 @default.
- W3037039100 creator A5030428283 @default.
- W3037039100 creator A5037162168 @default.
- W3037039100 creator A5078812767 @default.
- W3037039100 date "2020-06-03" @default.
- W3037039100 modified "2023-10-03" @default.
- W3037039100 title "Understanding Generalization in Deep Learning via Tensor Methods" @default.
- W3037039100 hasPublicationYear "2020" @default.
- W3037039100 type Work @default.
- W3037039100 sameAs 3037039100 @default.
- W3037039100 citedByCount "2" @default.
- W3037039100 countsByYear W30370391002021 @default.
- W3037039100 crossrefType "proceedings-article" @default.
- W3037039100 hasAuthorship W3037039100A5008267969 @default.
- W3037039100 hasAuthorship W3037039100A5010098535 @default.
- W3037039100 hasAuthorship W3037039100A5030428283 @default.
- W3037039100 hasAuthorship W3037039100A5037162168 @default.
- W3037039100 hasAuthorship W3037039100A5078812767 @default.
- W3037039100 hasConcept C104317684 @default.
- W3037039100 hasConcept C105795698 @default.
- W3037039100 hasConcept C108583219 @default.
- W3037039100 hasConcept C11413529 @default.
- W3037039100 hasConcept C119857082 @default.
- W3037039100 hasConcept C134306372 @default.
- W3037039100 hasConcept C154945302 @default.
- W3037039100 hasConcept C155281189 @default.
- W3037039100 hasConcept C177148314 @default.
- W3037039100 hasConcept C185592680 @default.
- W3037039100 hasConcept C202444582 @default.
- W3037039100 hasConcept C27158222 @default.
- W3037039100 hasConcept C33923547 @default.
- W3037039100 hasConcept C41008148 @default.
- W3037039100 hasConcept C50644808 @default.
- W3037039100 hasConcept C55493867 @default.
- W3037039100 hasConcept C63479239 @default.
- W3037039100 hasConceptScore W3037039100C104317684 @default.
- W3037039100 hasConceptScore W3037039100C105795698 @default.
- W3037039100 hasConceptScore W3037039100C108583219 @default.
- W3037039100 hasConceptScore W3037039100C11413529 @default.
- W3037039100 hasConceptScore W3037039100C119857082 @default.
- W3037039100 hasConceptScore W3037039100C134306372 @default.
- W3037039100 hasConceptScore W3037039100C154945302 @default.
- W3037039100 hasConceptScore W3037039100C155281189 @default.
- W3037039100 hasConceptScore W3037039100C177148314 @default.
- W3037039100 hasConceptScore W3037039100C185592680 @default.
- W3037039100 hasConceptScore W3037039100C202444582 @default.
- W3037039100 hasConceptScore W3037039100C27158222 @default.
- W3037039100 hasConceptScore W3037039100C33923547 @default.
- W3037039100 hasConceptScore W3037039100C41008148 @default.
- W3037039100 hasConceptScore W3037039100C50644808 @default.
- W3037039100 hasConceptScore W3037039100C55493867 @default.
- W3037039100 hasConceptScore W3037039100C63479239 @default.
- W3037039100 hasLocation W30370391001 @default.
- W3037039100 hasOpenAccess W3037039100 @default.
- W3037039100 hasPrimaryLocation W30370391001 @default.
- W3037039100 hasRelatedWork W2767449908 @default.
- W3037039100 hasRelatedWork W2898077722 @default.
- W3037039100 hasRelatedWork W2949579938 @default.
- W3037039100 hasRelatedWork W2962824341 @default.
- W3037039100 hasRelatedWork W2970337468 @default.
- W3037039100 hasRelatedWork W2996151541 @default.
- W3037039100 hasRelatedWork W2996510349 @default.
- W3037039100 hasRelatedWork W2999913149 @default.
- W3037039100 hasRelatedWork W3000569652 @default.
- W3037039100 hasRelatedWork W3022603917 @default.
- W3037039100 hasRelatedWork W3034891383 @default.
- W3037039100 hasRelatedWork W3086017325 @default.
- W3037039100 hasRelatedWork W3099532375 @default.
- W3037039100 hasRelatedWork W3102249364 @default.
- W3037039100 hasRelatedWork W3135482464 @default.
- W3037039100 hasRelatedWork W3162265811 @default.
- W3037039100 hasRelatedWork W3173999938 @default.
- W3037039100 hasRelatedWork W3182899221 @default.
- W3037039100 hasRelatedWork W3184961026 @default.
- W3037039100 hasRelatedWork W3096979466 @default.
- W3037039100 isParatext "false" @default.
- W3037039100 isRetracted "false" @default.
- W3037039100 magId "3037039100" @default.
- W3037039100 workType "article" @default.