Pattern Recognition and Machine Learning
Citations
205 citations
Cites background or methods from "Pattern Recognition and Machine Lea..."
...Since MoG is a universal approximator to any continuous probability distribution (Bishop, 2006; Meng & De la Torre, 2013), the proposed MoG-RPCA approach is capable of adapting a much wider range of real noises than the current RPCA methods....
[...]
...We use the variational Bayes (VB) (Bishop, 2006) method to infer the posterior of MoG-RPCA....
[...]
...Moreover, the rank needs to be pre-specified in this line of research, which is often unavailable in practice....
[...]
...To improve RPCA, a natural idea is to use MoG to model noise since MoG is a universal approximator to any continuous distributions (Bishop, 2006)....
[...]
...Setting of the Hyperparameters: We set all the hyperparameters involved in our model in a non-informative manner to make them influence as less as possible the inference of posterior distributions (Bishop, 2006)....
[...]
205 citations
205 citations
205 citations
Cites background or methods from "Pattern Recognition and Machine Lea..."
...In standard gradient descent, one computes the gradient of the objective loss function using all training examples, which is then used to adjust the parameter vector in the direction opposite to the gradient [6]....
[...]
...It is, of course, impossible to do justice to the immense literature on machine learning in the space available for this paper; for more details, we refer the reader to standard textbooks [6, 23]....
[...]
205 citations