.png){kind=icon}
.png){kind=icon}
Return
Magnitude Sensitive Competitive Learning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.neucom.2012.11.039.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper presents a new neural method for unsupervised learning denoted as Magnitude Sensitive Competitive Learning (MSCL), which has the property of distributing the unit centroids following any magnitude calculated from the unit parameters or the input data inside its Voronoi region. This controlled behavior permits it to outperform standard Competitive Learning algorithms that only tend to concentrate neurons according to the input data density, when other kind of data information processing is desired. Some examples applying different target functions show the MSCL possibilities in several applications as data-series interpolation, surface modelling from 3D point clouds and color quantization (CQ). (C) 2013 Elsevier B.V. All rights reserved.
Keywords:
Competitive learning
Neural Networks
Vector quantization
Surface modelling
Color quantization
Interpolation