Mining information from high dimensional and sparse datasets with statistical pattern recognition techniques . Palestrante: Dr. Carlos E. Thomaz (FEI)
Modern data acquisition methods are able to present the user with an impressive level of detail. As a consequence, automatic learning systems face ever increasing difficulty in separating the details relevant to a task from the myriad of available features. In this talk, I will present a couple of ideas of using multivariate statistical approaches to interpret and reconstruct data samples rather than only recognising them. These approaches address the difficulty of dealing with the inherent high dimensionality and limited sample size of several data analysis problems, and can be based on either a parametric or non-parametric pattern recognition strategy. To demonstrate the performance of these techniques I will show some experimental results on medical data compo