Batch and median neural gas
arXiv:math/0610561 · doi:10.1016/j.neunet.2006.05.018
Abstract
Neural Gas (NG) constitutes a very robust clustering algorithm given euclidian data which does not suffer from the problem of local minima like simple vector quantization, or topological restrictions like the self-organizing map. Based on the cost function of NG, we introduce a batch variant of NG which shows much faster convergence and which can be interpreted as an optimization of the cost function by the Newton method. This formulation has the additional benefit that, based on the notion of the generalized median in analogy to Median SOM, a variant for non-vectorial proximity data can be introduced. We prove convergence of batch and median versions of NG, SOM, and k-means in a unified formulation, and we investigate the behavior of the algorithms in several experiments.
In Special Issue after WSOM 05 Conference, 5-8 september, 2005, Paris
References in corpus (2)
Cited by in corpus (7)
- Median evidential c-means algorithm and its application to community detection
- Turbulence, Coherence and Collapse: Three Phases for Core Evolution
- Exploratory Analysis of Functional Data via Clustering and Optimal Segmentation
- How Many Dissimilarity/Kernel Self Organizing Map Variants Do We Need?
- Median topographic maps for biomedical data sets
- Multilinear objective function-based clustering
- Accélération des cartes auto-organisatrices sur tableau de dissimilarités par séparation et évaluation