Discriminative Robust Transformation Learning

9 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Robust Large Margin Deep Neural Networks (2016, background): “The work in [28] shows that learning transformations that are locally isometric is robust and leads to a small GE.”
Generalization Error of Invariant Classifiers (2016, background): “There are various works that that study the GE of CNNs (Huang et al., 2015; Neyshabur et al., 2015; Shalev-Shwartz and Ben-David, 2014; Sokolić et al., 2016), however, they do not establish any connection between the network’s invariance and its GE.”
Deep Neural Networks with Random Gaussian Weights: A Universal Classification Strategy? (2015, background): “Interestingly, follow-up studies [51], [52] found that adding metric preservation constraints to t he training of networks also leads to a theoretical relation be twe n the complexity of the data and the number of training samples .”
Mathematics of Deep Learning (2017, background): “An alternative route followed by [28], [29], [69], [73] bounds the generalization error in terms of the networks’ classiﬁcation margin, which is independent of the depth and size but takes into account the structure of the data (considering its covering number) and therefore avoids the above issues.”
Orthogonal Deep Neural Networks (2019, methodology): “To avoid this, [59] and [27] employ a large covering number to ensure that examples in the same ball have the same label.”