Matches in SemOpenAlex for { <https://semopenalex.org/work/W3215737631> ?p ?o ?g. }
Showing items 1 to 63 of
63
with 100 items per page.
- W3215737631 abstract "The field of Deep Learning is rich with empirical evidence of human-like performance on a variety of prediction tasks. However, despite these successes, the recent Predicting Generalization in Deep Learning (PGDL) NeurIPS 2020 competition suggests that there is a need for more robust and efficient measures of network generalization. In this work, we propose a new framework for evaluating the generalization capabilities of trained networks. We use perturbation response (PR) curves that capture the accuracy change of a given network as a function of varying levels of training sample perturbation. From these PR curves, we derive novel statistics that capture generalization capability. Specifically, we introduce two new measures for accurately predicting generalization gaps: the Gi-score and Pal-score, which are inspired by the Gini coefficient and Palma ratio (measures of income inequality), that accurately predict generalization gaps. Using our framework applied to intra and inter-class sample mixup, we attain better predictive scores than the current state-of-the-art measures on a majority of tasks in the PGDL competition. In addition, we show that our framework and the proposed statistics can be used to capture to what extent a trained network is invariant to a given parametric input transformation, such as rotation or translation. Therefore, these generalization gap prediction statistics also provide a useful means for selecting optimal network architectures and hyperparameters that are invariant to a certain perturbation." @default.
- W3215737631 created "2021-12-06" @default.
- W3215737631 creator A5024311601 @default.
- W3215737631 creator A5035487923 @default.
- W3215737631 creator A5050344371 @default.
- W3215737631 creator A5064335115 @default.
- W3215737631 date "2021-06-08" @default.
- W3215737631 modified "2023-09-23" @default.
- W3215737631 title "Predicting Deep Neural Network Generalization with Perturbation Response Curves" @default.
- W3215737631 doi "https://doi.org/10.48550/arxiv.2106.04765" @default.
- W3215737631 hasPublicationYear "2021" @default.
- W3215737631 type Work @default.
- W3215737631 sameAs 3215737631 @default.
- W3215737631 citedByCount "0" @default.
- W3215737631 crossrefType "posted-content" @default.
- W3215737631 hasAuthorship W3215737631A5024311601 @default.
- W3215737631 hasAuthorship W3215737631A5035487923 @default.
- W3215737631 hasAuthorship W3215737631A5050344371 @default.
- W3215737631 hasAuthorship W3215737631A5064335115 @default.
- W3215737631 hasBestOaLocation W32157376311 @default.
- W3215737631 hasConcept C119857082 @default.
- W3215737631 hasConcept C121332964 @default.
- W3215737631 hasConcept C134306372 @default.
- W3215737631 hasConcept C154945302 @default.
- W3215737631 hasConcept C177148314 @default.
- W3215737631 hasConcept C177918212 @default.
- W3215737631 hasConcept C190470478 @default.
- W3215737631 hasConcept C33923547 @default.
- W3215737631 hasConcept C37914503 @default.
- W3215737631 hasConcept C41008148 @default.
- W3215737631 hasConcept C50644808 @default.
- W3215737631 hasConcept C62520636 @default.
- W3215737631 hasConcept C8642999 @default.
- W3215737631 hasConceptScore W3215737631C119857082 @default.
- W3215737631 hasConceptScore W3215737631C121332964 @default.
- W3215737631 hasConceptScore W3215737631C134306372 @default.
- W3215737631 hasConceptScore W3215737631C154945302 @default.
- W3215737631 hasConceptScore W3215737631C177148314 @default.
- W3215737631 hasConceptScore W3215737631C177918212 @default.
- W3215737631 hasConceptScore W3215737631C190470478 @default.
- W3215737631 hasConceptScore W3215737631C33923547 @default.
- W3215737631 hasConceptScore W3215737631C37914503 @default.
- W3215737631 hasConceptScore W3215737631C41008148 @default.
- W3215737631 hasConceptScore W3215737631C50644808 @default.
- W3215737631 hasConceptScore W3215737631C62520636 @default.
- W3215737631 hasConceptScore W3215737631C8642999 @default.
- W3215737631 hasLocation W32157376311 @default.
- W3215737631 hasOpenAccess W3215737631 @default.
- W3215737631 hasPrimaryLocation W32157376311 @default.
- W3215737631 hasRelatedWork W2804162248 @default.
- W3215737631 hasRelatedWork W3199608561 @default.
- W3215737631 hasRelatedWork W4210794429 @default.
- W3215737631 hasRelatedWork W4223456145 @default.
- W3215737631 hasRelatedWork W4280535922 @default.
- W3215737631 hasRelatedWork W4294564511 @default.
- W3215737631 hasRelatedWork W4295309597 @default.
- W3215737631 hasRelatedWork W4309113015 @default.
- W3215737631 hasRelatedWork W4313854490 @default.
- W3215737631 hasRelatedWork W1629725936 @default.
- W3215737631 isParatext "false" @default.
- W3215737631 isRetracted "false" @default.
- W3215737631 magId "3215737631" @default.
- W3215737631 workType "article" @default.