Multilayer Nonnegative Matrix Factorization Using Projected Gradient Approaches, paper

Jie Fu https://sites.google.com/site/bigaidream/ Keywords: NMF, Hierarchical

Introduction

[Open Issues] The performance of many existing NMF algorithms may be quite poor, especially, when the unknown nonnegative components are badly scaled (ill-conditioned data), insufficiently sparse, and a number of observations is equal or only slightly greater than a number of latent components. [Solution] the authors developed a simple hierarchical and multistage {deep learning?} procedure in which they perform a sequential decomposition of nonnegative matrices as follows: 1. Perform the basic decomposition X=A1S1 using any available NMF algorithm. 2. The results obtained from the first stage are used to perform the similar decomposition S1=A2S2 using the same or different update rules, and so on. The model has the form: X=A1A2ALSL, with the basis matrix defined as A=A1A2AL. Physically, this means that we build up a system that has many layers or cascade connection of L mixing subsystems. The key point is that the learning (update) process to find parameters of sub-matrices Al and Sl is performed sequentially, i.e., layer by layer and in each layer we use multi-start initialization. The toolbox can be found here: http://www.bsp.brain.riken.jp/index.php