Self-organizing maps
GTM: the generative topographic mapping
Neural Computation
Learning in graphical models
The Cross Entropy Method: A Unified Approach To Combinatorial Optimization, Monte-carlo Simulation (Information Science and Statistics)
Proceedings of the 34th conference on Winter simulation: exploring new frontiers
A fast learning algorithm for deep belief nets
Neural Computation
Two topographic maps for data visualisation
Data Mining and Knowledge Discovery
Multi-layer topology preserving mapping for K-means clustering
IDEAL'11 Proceedings of the 12th international conference on Intelligent data engineering and automated learning
| Hi-index | 0.00 |
In this paper, we review the technique of Cross Entropy and apply it in a novel manner to the creation of topology preserving mappings. We create an underlying latent space which is mapped into data space by mapping the latent points through a number of basis functions and then linearly combining the output of these basis functions to create centres or prototypes in data space. In this paper, we use the cross entropy method for adapting the parameters of this linear combination matrix. We further show that the method can be extended to deep architectures, architectures with multiple layers of adaptive units which can be used to try to create more interesting features than can be found with one adaptive layer.