Generative Topographic Mapping (GTM) is a non-linear latent variable model that provides simultaneous visualization and clustering of high-dimensional data. It was originally formulated as a constrained mixture of distributions, for which the adaptive parameters were determined by Maximum Likelihood (ML), using the Expectation-Maximization (EM) algorithm. In this paper, we define an alternative variational formulation of GTM that provides a full Bayesian treatment to a Gaussian Process (GP)-based variation of GTM. The performance of the proposed Variational GTM is assessed in several experiments with artificial datasets. These experiments highlight the capability of Variational GTM to avoid data overfitting through active regularization.
KeywordsLatent Space Marginal Likelihood Adaptive Parameter Manifold Learning Generative Topographic Mapping
Unable to display preview. Download preview PDF.
- 5.Beal, M.: Variational algorithms for approximate Bayesian inference. PhD thesis, The Gatsby Computational Neuroscience Unit, Univ. College London (2003)Google Scholar
- 7.Abrahamsen, P.: A review of Gaussian random fields and correlation functions. Technical Report 917, Norwegian Computing Center, Oslo, Norway (1997)Google Scholar
- 9.Bishop, C.M.: Variational principal components. In: Proceedings Ninth Intern. Conf. on Artificial Neural Networks, vol. 1, pp. 509–514 (1999)Google Scholar
- 12.Olier, I., Vellido, A.: A variational Bayesian formulation for GTM: Theoretical foundations. Technical report, Technical University of Catalonia (UPC) (2007)Google Scholar