Texture Modeling with Convolutional Spike-and-Slab RBMs and Deep Extensions

By Heng Luo, Pierre Luc Carrier, Aaron C. Courville, Yoshua Bengio. 2012

25 citations recorded by the source corpus.

We apply the spike-and-slab Restricted Boltzmann Machine (ssRBM) to texture modeling. The ssRBM with tiled-convolution weight sharing (TssRBM) achieves or surpasses the state-of-the-art on texture synthesis and inpainting by parametric models. We also develop a novel RBM model with a spike-and-slab visible layer and binary variables in the hidden layer. This model is designed to be stacked on top of the TssRBM. We show the resulting deep belief network (DBN) is a powerful generative model that improves on single-layer models and is capable of modeling not only single high-resolution and challenging textures but also multiple textures.

Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, Signal Processing

How researchers see this paper

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

In Computer Graphics and Computer-Aided Design

1. Yoshua Bengio
2. Richard S. Zemel
3. Kyunghyun Cho

In Computer Vision and Pattern Recognition

1. Yoshua Bengio
2. Luc Van Gool
3. Dacheng Tao
4. Ming-Hsuan Yang
5. Trevor Darrell
6. Jitendra Malik
7. Andrea Vedaldi
8. Xiaogang Wang

In Signal Processing

1. Yoshua Bengio
2. Kyunghyun Cho
3. Tapani Raiko
4. Richard S. Zemel
5. Daniel Tarlow
6. Jason Yosinski
7. Guillaume Alain
8. Eric Thibodeau-Laufer

Citations in the corpus

• Deep Generative Stochastic Networks Trainable by Backprop (2013, methodology): “Gaussian mixture rather than a Bernoulli mixture to compute the likelihood, but we can compare with Rifai et al. (2012); Bengio et al.”
• Deep Learning of Representations: Looking Forward (2013, background): “. (2010); Cho et al. (2010); Salakhutdinov (2010b,a) have 18 See examples of generated images with some of the current state-of-the-art in learned generative models of images (Courville et al., 2011; Luo et al., 2013) all considered various forms of tempering to address the failure of Gibbs chain mi”
• A Two-Stage Pretraining Algorithm for Deep Boltzmann Machines (2013, background): “Any other machine learning model that gives reasonable posterior distributions over multiple layers of binary hidden units can be used instead of RBMs or DAEs.”
• Exploring Compositional High Order Pattern Potentials for Structured Output Learning (2013, methodology): “A few methods developped for tiled convolutional (not fully convolutional) RBMs achieved good results modeling textures [8][19], which shows some potential that this may be a good way to go.”