Matches in SemOpenAlex for { <https://semopenalex.org/work/W4310698909> ?p ?o ?g. }
Showing items 1 to 68 of
68
with 100 items per page.
- W4310698909 abstract "Deep neural networks (DNN) have been widely applied in modern life, including critical domains like autonomous driving, making it essential to ensure the reliability and robustness of DNN-powered systems. As an analogy to code coverage metrics for testing conventional software, researchers have proposed neuron coverage metrics and coverage-driven methods to generate DNN test cases. However, Yan et al. doubt the usefulness of existing coverage criteria in DNN testing. They show that a coverage-driven method is less effective than a gradient-based method in terms of both uncovering defects and improving model robustness. In this paper, we conduct a replication study of the work by Yan et al. and extend the experiments for deeper analysis. A larger model and a dataset of higher resolution images are included to examine the generalizability of the results. We also extend the experiments with more test case generation techniques and adjust the process of improving model robustness to be closer to the practical life cycle of DNN development. Our experiment results confirm the conclusion from Yan et al. that coverage-driven methods are less effective than gradient-based methods. Yan et al. find that using gradient-based methods to retrain cannot repair defects uncovered by coverage-driven methods. They attribute this to the fact that the two types of methods use different perturbation strategies: gradient-based methods perform differentiable transformations while coverage-driven methods can perform additional non-differentiable transformations. We test several hypotheses and further show that even coverage-driven methods are constrained only to perform differentiable transformations, the uncovered defects still cannot be repaired by adversarial training with gradient-based methods. Thus, defensive strategies for coverage-driven methods should be further studied." @default.
- W4310698909 created "2022-12-16" @default.
- W4310698909 creator A5002667771 @default.
- W4310698909 creator A5022802322 @default.
- W4310698909 creator A5049635019 @default.
- W4310698909 creator A5081036622 @default.
- W4310698909 date "2022-01-01" @default.
- W4310698909 modified "2023-10-03" @default.
- W4310698909 title "Revisiting Neuron Coverage Metrics and Quality of Deep Neural Networks" @default.
- W4310698909 doi "https://doi.org/10.48550/arxiv.2201.00191" @default.
- W4310698909 hasPublicationYear "2022" @default.
- W4310698909 type Work @default.
- W4310698909 citedByCount "0" @default.
- W4310698909 crossrefType "posted-content" @default.
- W4310698909 hasAuthorship W4310698909A5002667771 @default.
- W4310698909 hasAuthorship W4310698909A5022802322 @default.
- W4310698909 hasAuthorship W4310698909A5049635019 @default.
- W4310698909 hasAuthorship W4310698909A5081036622 @default.
- W4310698909 hasBestOaLocation W43106989091 @default.
- W4310698909 hasConcept C104317684 @default.
- W4310698909 hasConcept C105795698 @default.
- W4310698909 hasConcept C119857082 @default.
- W4310698909 hasConcept C134306372 @default.
- W4310698909 hasConcept C137726913 @default.
- W4310698909 hasConcept C154945302 @default.
- W4310698909 hasConcept C185592680 @default.
- W4310698909 hasConcept C202615002 @default.
- W4310698909 hasConcept C27158222 @default.
- W4310698909 hasConcept C2984842247 @default.
- W4310698909 hasConcept C33923547 @default.
- W4310698909 hasConcept C41008148 @default.
- W4310698909 hasConcept C50644808 @default.
- W4310698909 hasConcept C55493867 @default.
- W4310698909 hasConcept C58166 @default.
- W4310698909 hasConcept C63479239 @default.
- W4310698909 hasConceptScore W4310698909C104317684 @default.
- W4310698909 hasConceptScore W4310698909C105795698 @default.
- W4310698909 hasConceptScore W4310698909C119857082 @default.
- W4310698909 hasConceptScore W4310698909C134306372 @default.
- W4310698909 hasConceptScore W4310698909C137726913 @default.
- W4310698909 hasConceptScore W4310698909C154945302 @default.
- W4310698909 hasConceptScore W4310698909C185592680 @default.
- W4310698909 hasConceptScore W4310698909C202615002 @default.
- W4310698909 hasConceptScore W4310698909C27158222 @default.
- W4310698909 hasConceptScore W4310698909C2984842247 @default.
- W4310698909 hasConceptScore W4310698909C33923547 @default.
- W4310698909 hasConceptScore W4310698909C41008148 @default.
- W4310698909 hasConceptScore W4310698909C50644808 @default.
- W4310698909 hasConceptScore W4310698909C55493867 @default.
- W4310698909 hasConceptScore W4310698909C58166 @default.
- W4310698909 hasConceptScore W4310698909C63479239 @default.
- W4310698909 hasLocation W43106989091 @default.
- W4310698909 hasLocation W43106989092 @default.
- W4310698909 hasOpenAccess W4310698909 @default.
- W4310698909 hasPrimaryLocation W43106989091 @default.
- W4310698909 hasRelatedWork W2952999927 @default.
- W4310698909 hasRelatedWork W2962182036 @default.
- W4310698909 hasRelatedWork W3038036429 @default.
- W4310698909 hasRelatedWork W4200511449 @default.
- W4310698909 hasRelatedWork W4225555082 @default.
- W4310698909 hasRelatedWork W4300978433 @default.
- W4310698909 hasRelatedWork W4308233811 @default.
- W4310698909 hasRelatedWork W4311734044 @default.
- W4310698909 hasRelatedWork W4313044179 @default.
- W4310698909 hasRelatedWork W4328028516 @default.
- W4310698909 isParatext "false" @default.
- W4310698909 isRetracted "false" @default.
- W4310698909 workType "article" @default.