Corpus ID: 227746659

Rotation-Invariant Autoencoders for Signals on Spheres

@article{Lohit2020RotationInvariantAF,
  title={Rotation-Invariant Autoencoders for Signals on Spheres},
  author={Suhas Lohit and Shubhendu Trivedi},
  journal={ArXiv},
  year={2020},
  volume={abs/2012.04474}
}
Omnidirectional images and spherical representations of $3D$ shapes cannot be processed with conventional 2D convolutional neural networks (CNNs) as the unwrapping leads to large distortion. Using fast implementations of spherical and $SO(3)$ convolutions, researchers have recently developed deep learning methods better suited for classifying spherical images. These newly proposed convolutional layers naturally extend the notion of convolution to functions on the unit sphere $S^2$ and the group… Expand

Figures and Tables from this paper

You Only Hypothesize Once: Point Cloud Registration with Rotation-equivariant Descriptors
TLDR
A novel local descriptor-based framework, called You Only Hypothesize Once (YOHO), for the registration of two unaligned point clouds, which achieves the rotation invariance by recent technologies of group equivariant feature learning, which brings more robustness to point density and noise. Expand

References

SHOWING 1-10 OF 50 REFERENCES
Learning SO(3) Equivariant Representations with Spherical CNNs
TLDR
It is shown that networks with much lower capacity and without requiring data augmentation can exhibit performance comparable to the state of the art in standard retrieval and classification benchmarks. Expand
Spherical CNNs
TLDR
A definition for the spherical cross-correlation is proposed that is both expressive and rotation-equivariant and satisfies a generalized Fourier theorem, which allows us to compute it efficiently using a generalized (non-commutative) Fast Fourier Transform (FFT) algorithm. Expand
Harmonic Networks: Deep Translation and Rotation Equivariance
TLDR
H-Nets are presented, a CNN exhibiting equivariance to patch-wise translation and 360-rotation, and it is demonstrated that their layers are general enough to be used in conjunction with the latest architectures and techniques, such as deep supervision and batch normalization. Expand
Learning Steerable Filters for Rotation Equivariant CNNs
TLDR
Steerable Filter CNNs (SFCNNs) are developed which achieve joint equivariance under translations and rotations by design and generalize He's weight initialization scheme to filters which are defined as a linear combination of a system of atomic filters. Expand
Towards Learning Affine-Invariant Representations via Data-Efficient CNNs
TLDR
A novel multi-scale maxout CNN is proposed and train it end-to-end with a novel rotation-invariant regularizer that aims to enforce the weights in each 2D spatial filter to approximate circular patterns. Expand
Extracting Invariant Features From Images Using An Equivariant Autoencoder
TLDR
This work applies group convolutions to build an Equivariant Autoencoder with embeddings that change predictably under the specified set of transformations, and introduces two approaches to extracting invariant features from theseembeddings—Gram Pooling and Equivariants Attention. Expand
Gauge Equivariant Convolutional Networks and the Icosahedral CNN
TLDR
Gauge equivariant convolution using a single conv2d call is demonstrated, making it a highly scalable and practical alternative to Spherical CNNs and demonstrating substantial improvements over previous methods on the task of segmenting omnidirectional images and global climate patterns. Expand
Measuring Invariances in Deep Networks
TLDR
A number of empirical tests are proposed that directly measure the degree to which these learned features are invariant to different input transformations and find that stacked autoencoders learn modestly increasingly invariant features with depth when trained on natural images and convolutional deep belief networks learn substantially more invariant Features in each layer. Expand
Deforming Autoencoders: Unsupervised Disentangling of Shape and Appearance
TLDR
A more powerful form of unsupervised disentangling becomes possible in template coordinates, allowing us to successfully decompose face images into shading and albedo, and further manipulate face images. Expand
Deep Convolutional Inverse Graphics Network
This paper presents the Deep Convolution Inverse Graphics Network (DC-IGN), a model that aims to learn an interpretable representation of images, disentangled with respect to three-dimensional sceneExpand
...
1
2
3
4
5
...