2019
DOI: 10.1109/tpami.2018.2855738
|View full text |Cite
|
Sign up to set email alerts
|

Scattering Networks for Hybrid Representation Learning

Abstract: Scattering networks are a class of designed Convolutional Neural Networks (CNNs) with fixed weights. We argue they can serve as generic representations for modelling images. In particular, by working in scattering space, we achieve competitive results both for supervised and unsupervised learning tasks, while making progress towards constructing more interpretable CNNs. For supervised learning, we demonstrate that the early layers of CNNs do not necessarily need to be learned, and can be replaced with a scatte… Show more

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

1
66
0

Year Published

2019
2019
2023
2023

Publication Types

Select...
5
1
1

Relationship

0
7

Authors

Journals

citations
Cited by 67 publications
(67 citation statements)
references
References 27 publications
1
66
0
Order By: Relevance
“…To further demonstrate the superiority of CapGAN in generating images for classification task, we have performed the binary classification experiment on the real prostate cancer dataset via the traditional augmentation (TraAug), DCGAN-based augmentation (DCGANAug), LSGAN-based augmentation (LSGANAug), and CapGAN-based augmentation (CapGANAug) methods. Recent studies [ 43 ] have demonstrated that the early layers of deep learning models can be replaced with predefined representations to reduce and parametrize variability while retaining the discriminative information. Inspired by this motivation, we have proposed to combine feature representation with the micro networks for prostate MR image classification.…”
Section: Resultsmentioning
confidence: 99%
“…To further demonstrate the superiority of CapGAN in generating images for classification task, we have performed the binary classification experiment on the real prostate cancer dataset via the traditional augmentation (TraAug), DCGAN-based augmentation (DCGANAug), LSGAN-based augmentation (LSGANAug), and CapGAN-based augmentation (CapGANAug) methods. Recent studies [ 43 ] have demonstrated that the early layers of deep learning models can be replaced with predefined representations to reduce and parametrize variability while retaining the discriminative information. Inspired by this motivation, we have proposed to combine feature representation with the micro networks for prostate MR image classification.…”
Section: Resultsmentioning
confidence: 99%
“…compression [1]. This paper extends further the proposed harmonic block by: 1) showing how it relates to the modified discrete cosine transform when considering overlap in computing convolution, 2) proposing an improved, computationally more efficient implementation, and 3) showing that the CNNs using the harmonic block outperform scattering network, based on the use of wavelet-based filters [2], [3] when training data is scarce. The PyTorch implementation of the harmonic block is provided at https:// github.com/ matej-ulicny/ harmonic-networks.…”
Section: Introductionmentioning
confidence: 84%
“…Silva et al used wavelet filters to enhance edges prior to CNN processing [15]. Rotation and scale invariant wavelet based scattering networks with subsequent CNN were formulated in [3], [16]. These hybrid networks were shown to reach comparable classification accuracy to deeper CNNs.…”
Section: B Wavelets and Cnnsmentioning
confidence: 99%
See 2 more Smart Citations