Corpus ID: 21959489

Improving Content-Invariance in Gated Autoencoders for 2D and 3D Object Rotation

@article{Lattner2017ImprovingCI,
  title={Improving Content-Invariance in Gated Autoencoders for 2D and 3D Object Rotation},
  author={S. Lattner and M. Grachten},
  journal={ArXiv},
  year={2017},
  volume={abs/1707.01357}
}
  • S. Lattner, M. Grachten
  • Published 2017
  • Computer Science
  • ArXiv
  • Content-invariance in mapping codes learned by GAEs is a useful feature for various relation learning tasks. In this paper we show that the content-invariance of mapping codes for images of 2D and 3D rotated objects can be substantially improved by extending the standard GAE loss (symmetric reconstruction error) with a regularization term that penalizes the symmetric cross-reconstruction error. This error term involves reconstruction of pairs with mapping codes obtained from other pairs… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    Explore Further: Topics Discussed in This Paper

    Learning Musical Relations using Gated Autoencoders
    Modeling Musical Structure with Artificial Neural Networks

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 33 REFERENCES
    Unsupervised Learning of Image Transformations
    197
    Learning to Represent Spatial Transformations with Factored Higher-Order Boltzmann Machines
    245
    Gradient-based learning of higher-order image features
    61
    On multi-view feature learning
    48
    Bilinear models of natural images
    49
    Modeling the joint density of two images under a variety of transformations
    58
    Learning to Disentangle Factors of Variation with Manifold Interaction
    177
    Emergence of Phase- and Shift-Invariant Features by Decomposition of Natural Images into Independent Feature Subspaces
    585