Hyperparameter selection for self-organizing maps
Neural Computation
Self-organizing maps
GTM: the generative topographic mapping
Neural Computation
S-map: a network with a simple self-organization algorithm for generative topographic mappings
NIPS '97 Proceedings of the 1997 conference on Advances in neural information processing systems 10
Neural maps and topographic vector quantization
Neural Networks
Bayesian Sampling and Ensemble Learning in Generative Topographic Mapping
Neural Processing Letters
Density Estimation by Mixture Models with Smoothing Priors
Neural Computation
Self organization of a massive document collection
IEEE Transactions on Neural Networks
Self-Organising maps for classification with metropolis-hastings algorithm for supervision
ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part III
Hi-index | 0.00 |
As data sets get larger and larger, the need for exploratory methods that allow some visualization of the overall structure in the data is becoming more important. The self-organizing map (SOM) introduced by Kohonen is a powerful tool for precisely this purpose. In recent years, SOM-based methodology has been refined and deployed with success in various high-dimensional problems. Still, our understanding of the properties of SOMs fitted by Kohonen's original algorithm is not complete, and several statistical models and alternative fitting algorithms have been devised in the literature. This paper presents a new Metropolis-Hastings Markov chain Monte Carlo algorithm designed for SOM fitting. The method stems from both the previous success of bayesian machinery in neural models and the uprise of computer-intensive, simulation-based algorithms in bayesian inference. Experimental results suggest the feasibility as well as the limitations of the approach in its current form. Since the method is based on a few extremely simple chain transition kernels, the framework may well accommodate the more sophisticated constructs needed for a full emulation of the self-organization treat.