I am interested in Machine Learning and Statistical Learning algorithms
for pattern analysis. My research focuses on their application within different
fields as insurance, bank, marketing, customer intelligence, environmental sciences,
biomedicine… During the last few years I have been involved in the internals of
Random Forests learning mechanism, for which I am
conducting research to develop new facilities and
utilities.
An in fashion domain where data mining has found challenging
applications is bioinformatics, where high throughput technologies have
provided the scenario for harvesting large amounts of biological and genetic
data. Classical procedures are not well suited for the analysis of these data
sets; hence, new research must be developed to deal with them. My research is
concerned with gene expression and proteomics data analysis, where one of the
main issues is the identification of hidden genomic patterns with biological
and medical implications that shed light on the internal molecular mechanisms
involved in diseases. The main feature of this type of data is its high
dimensional - low sample structure, which makes the analysis difficult and
cumbersome. I am developing algorithms and procedures able to extract relevant
and useful information from these data sets, which are expected to point out to
new biological findings and medical advances. I have been involved with mRNAs
expression data in breast cancer as well as with proteins and gene expression
data in bladder and colon cancers.
My
focus is on the study of flexible models that account for non-normality in many
real life applications where the data at hand does not fit a normal
distribution. I am interested both in the theory of multivariate non-normal
models, like skew-normal, elliptical or skew-elliptical distributions, as well
as in their applications for data modeling in real life problems.
Perhaps, this is the most theoretical branch of my research. It traces
back to the foundations of asymptotic statistics. I am quite interested in saddlepoint approximations, which was the topic of my Ph.D Thesis. I have explored the connection between
Edgeworth expansions and saddlepoint approximations,
as well as the link between the inversion of the former ---i.e. the
Cornish-Fisher approximations--- and the inversion of the latter.