Inducing wavelets into random fields via generative boosting

By Jianwen Xie, Yang Lu, Song-Chun Zhu, Ying Nian Wu. 2016

23 citations recorded by the source corpus.

This paper proposes a learning algorithm for the random field models whose energy functions are in the form of linear combinations of rectified filter responses from subsets of wavelets selected from a given over-complete dictionary. The algorithm consists of the following two components. (1) We propose to induce the wavelets into the random field model by a generative version of the epsilon-boosting algorithm. (2) We propose to generate the synthesized images from the random field model using Gibbs sampling on the coefficients (or responses) of the selected wavelets. We show that the proposed learning and sampling algorithms are capable of generating realistic image patterns. We also evaluate our learning method on a dataset of clustering tasks to demonstrate that the models can be learned in an unsupervised setting. The learned models encode the patterns in wavelet sparse coding. Moreover, they can be mapped to the second-layer nodes of a sparsely connected convolutional neural network (CNN).

Subfields: Oceanography, Generative Models, 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 Oceanography

1. Song-Chun Zhu
2. Ying Nian Wu
3. Jianwen Xie

In Generative Models

1. Song-Chun Zhu
2. Ying Nian Wu
3. Jianwen Xie
4. Ruiqi Gao
5. Yang Lu
6. Erik Nijkamp
7. Zilong Zheng
8. Wenguan Wang

In Signal Processing

1. Song-Chun Zhu
2. Ying Nian Wu
3. Jianwen Xie
4. Yang Lu
5. Ruiqi Gao
6. Wenguan Wang
7. Myle Ott
8. Arthur Szlam

Citations in the corpus

• Learning Energy-Based Spatial-Temporal Generative ConvNets for Dynamic Patterns (2019, methodology): “Our previous works [34] and [35] have studied this issue.”
• Learning Descriptor Networks for 3D Shape Synthesis and Analysis (2018, methodology): “The sparse FRAME model [30, 32], which was used for modeling object patterns.”
• A Theory of Generative ConvNet (2016, background): “Sparsifying the connections will make the compositions more explicit and meaningful (Xie et al., 2015; Wu et al., 2010).”
• Learning FRAME Models Using CNN Filters (2015, methodology): “In the paper of Xie et al. (2015a) , they use h ( r ) = | r | for full wave rectification.”