Products of Gaussians and Probabilistic Minor Component Analysis
By Christopher K. I. Williams, Felix V. Agakov. 2002
46 citations recorded by the source corpus.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Statistics and Probability
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Yoshua Bengio
- Ruslan Salakhutdinov
- Sergey Levine
- Quoc V. Le
- Kyunghyun Cho
- Aaron C. Courville
- Mohit Bansal
- Trevor Darrell
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Yoshua Bengio
- Dacheng Tao
- Trevor Darrell
- Ming-Hsuan Yang
- Xiaogang Wang
- Jitendra Malik
- Andrea Vedaldi
In Statistics and Probability
- Yoshua Bengio
- Aaron C. Courville
- Nicolas Manfred Otto Heess
- Andriy Mnih
- Geoffrey E. Hinton
- Zoubin Ghahramani
- William T. Freeman
- John P. Cunningham
Citations in the corpus
- How to Represent Part-Whole Hierarchies in a Neural Network (2021, methodology): “We could use a population of N M neurons each of which is tuned to a Gaussian in the M -dimensional space [Williams and Agakov, 2002].”
- Linear dimensionality reduction: survey, insights, and generalizations (2014, methodology): “Second, there have been several probabilistic extensions to PCA, such as probabilistic PCA (PPCA, (Tipping and Bishop, 1999; Roweis, 1997)), extreme component analysis (XCA, (Welling et al., 2003)), and minor component analysis (MCA, (Williams and Agakov, 2002)).”
- A Spike and Slab Restricted Boltzmann Machine (2011, methodology): “We also note that the covariance structure of the ssRBM conditional p(v | h) (equation 3) is essentially identical to the product of probabilistic principal components analysis (PoPPCA) model [27] with components corresponding to the ssRBM weight vectors associated with the active hidden units (hi =”
- Learning generative models of mid-level structure in natural images (2012, background): “They start by analyzing a Gaussian PoE and demonstrate that here the expected filters correspond to the minor components of the data (Williams and Agakov, 2002) and are thus expected to have high-frequency content.”
- Learning nonlinear constraints with contrastive backpropagation (2005, methodology): “Other datasets are better characterized by representing constraints that correspond to ways in which the data does not vary [Williams and Agakov, 2002].”
- What makes a good model of natural images? (2007, methodology): “We start by reviewing the results of Williams and Agakov [27] for Gaussian potentials.”