In applied statistics, our focus is in forest biometrics and in analysis of grouped, spatially and temporally dependent data. In forest biometrics, we work on applications of spatial point process theory and stochastic geometry in forest inventories. In analysis of dependent data, one major application is the modeling of greenhouse gas fluxes on peatlands based on chamber measurements. In machine learning, our focus is in statistical modeling from massive datasets, where typical data set size is 0.5TB. General goal is to estimate a generalizable model with which recognition can be performed on the previously unseen dataset. Previously, the group focused on recognition tasks from the speech signal, such as automatic speaker and language recognition. Recently, we have used image, video, text and molecule biological datasets, in addition to speech data.