F. Fortuna, F. Maturo, A. M. Aguilera, T. Di Battista
Extracting knowledge from high-dimensional data is a challenging task especially due to the curse of dimensionality. To address this issue, a two-step dimensionality reduction procedure based on functional data analysis is proposed. In the first step, high-dimensional data are divided into non-overlapping time-based windows and, for each of them, a probability density function is estimated to capture the main characteristics of the data. In the second step, functional principal component analysis is applied to some transformations of densities, which ensure their belonging to appropriate functional spaces. Specifically, the problem of forecasting probability density functions in new time windows is considered, taking into account the temporal dependence of the functional observations.
Palabras clave: high-dimensional data, dimensionality reduction, probability density functions, functional principal component analysis, functional time series
Programado
Sesión Invitada Hispano-Italiana
9 de junio de 2022 10:10
Sala de Conferencias
