Explore chapters and articles related to this topic
Introduction
Published in Kim-Hui Yap, Ling Guan, Stuart William Perry, Hau-San Wung, Adaptive Image Processing, 2018
Kim-Hui Yap, Ling Guan, Stuart William Perry, Hau-San Wung
The hierarchical feature map (HFM) [79] extends such ideas to a pyramidal hierarchy of SOFMs, each progressively trained in a top-down manner, to achieve some semblance of hierarchical partitioning on the input space. At the other end of the self-organizing spectrum is neural gas (NG) [80], which completely abandons any imposed topology: instead relying on the consideration of k nearest neighbors for the refinement of prototypes.
A Study on Behaviour of Neural Gas on Images and Artificial Neural Network in Healthcare
Published in Ashish Mishra, G. Suseendran, Trung-Nghia Phung, Soft Computing Applications and Techniques in Healthcare, 2020
Rahul Sahu, Ashish Mishra, G. Suseendran
Neural gas is a biologically adaptive algorithm, promoted by Martinetz and Schulten, in 1991. It arranges the input signals with relation to how much distance they have maintained. A specific quantity of these are suggested by distance, which is in some ratio; then many adaptation units and strength are dropped according to a fixed state.
Self-Organising Maps: The Hybrid SOM–NG Algorithm
Published in Qurban A. Memon, Distributed Networks, 2017
Mario J. Crespo, Iván Machón, Hilario López, Jose Luis Calvo
Neural gas (NG) is an algorithm similar to k-means, but it has a cooperative learning rule based on a distance ranking in the multidimensional input space. This algorithm produces the optimum quantisation results as it has no other restrictions to place prototypes.
Region of interest-based adaptive segmentation for image compression using hybrid Jaya–Lion mathematical approach
Published in International Journal of Computers and Applications, 2021
B. P. Santosh Kumar, K. Venkata Ramanaiah
In 2005, Anke et al. [29] have utilized a new model that was on the basis of topology-preserving neural networks for compressing the medical image. The explained model was a novel image compression process that has distinguished itself in different ways. This was supplied to huge blocks of image and has represented improved probability distribution assessment models. Subsequently, they have applied an operation (transformation-based) in the decomposed image. The progression of quantization was done by ‘neural-gas’ network. In order to examine the superiority of proposed work, they have examined the proposed model over other models.