Guest Editorial: Special Issue on Computational Intelligence for Industrial Data Processing and Analysis

Dianhui Wang,Wen Yu,Tianyou Chai
DOI: https://doi.org/10.1016/j.neucom.2015.03.079
IF: 6
2015-01-01
Neurocomputing
Abstract:This paper deals with Principal Components Analysis (PCA) of data spread over a network where central coordination and synchronous communication between networking nodes are forbidden. We propose an asynchronous and decentralized PCA algorithm dedicated to large scale problems, where “large” simultaneously applies to dimensionality, number of observations and network size. It is based on the integration of a dimension reduction step into a gossip consensus protocol. Unlike other approaches, a straightforward dual formulation makes it suitable when observed dimensions are distributed. We theoretically show its equivalence with a centralized PCA under a low-rank assumption on training data. An experimental analysis reveals that it achieves a good accuracy with a reasonable communication cost even when the low-rank assumption is relaxed.
What problem does this paper attempt to address?