Dr Alfredo Kalaitzis

Position Postdoctoral research associate
Email(*) a.kalaitzis
Themes Computational Statistics, Multivariate and High Dimensional Data

* @ucl.ac.uk

Biographical Details

I'm a postdoc within the CSML centre at UCL. My current work on the VOLTNET project, Under the direction of Dr James Nelson, combines ideas from Machine Learning and Signal Processing to solve problems affecting the power distribution industry. One such class of problems involves the detection, analysis and localisation of faults on low-voltage electric power transmission networks. Previously at the same department, I spent one year working with Dr Ricardo Silva on sparse latent variable models for ordinal and discrete data. I completed my PhD in 2012 at the University of Sheffield, under the guidance of Professor Neil D. Lawrence, where my primary focus was in generalising probabilistic models of Principal Component Analysis and other models of structured covariance matrices.

Research Interests

  • PCA
  • Gaussian processes
  • Markov chain Monte Carlo
  • copulas, sparsity

Selected publications

  • Flexible sampling of discrete data correlations without the marginal distributions, A. Kalaitzis, R. Silva, NIPS 2013.
  • The Bigraphical Lasso, A. Kalaitzis, J. Lafferty, N.D. Lawrence, S Zhou, ICML 2013.
  • Residual Component Analysis: Generalising PCA for more flexible inference in linear-Gaussian models, A. Kalaitzis, N.D. Lawrence, ICML 2012.
  • A simple approach to ranking differentially expressed gene expression time courses through gaussian process regression, A. Kalaitzis, N.D. Lawrence, BMC bioinformatics 12 (1), 180

Page last modified on 13 feb 14 14:45