Nonlinear mapping using particle swarm optimisation

Abstract— Nonlinear mapping is an approach of multidimensional scaling where a high-dimensional space is transformed into a lower-dimensional space such that the topological characteristics of the original high-dimensional space are preserved. This enables visualisation and feature extraction of datasets. Problems exist in conventional mapping methods in that they can be slow or illustrate poor performance. Particle Swarm Optimisation (PSO) is a global optimisation approach that can effectively and efficiently map vectors from one dimension to another. A lowdimensional representation of vectors is used for classification purposes applicable in a number of security applications, such as face and hand recognition. Multimedia is regularly transmitted across the network, and needs to be protected. Watermarking has been designed to be such a protection scheme. The weakness of existing watermark schemes is briefly discussed. The ability to estimate the lower-dimensional watermarking sub...

• Type: Conference
• Year: 2005
• Where: CEC
• Authors: Auralia I. Edwards, Andries Petrus Engelbrecht, Nelis Franken
• Keywords: Artificial Intelligence, CEC 2005, Conventional Mapping Methods, Lower-dimensional Watermarking Subspace, Mapping Method