A recurrent neural network (RNN) is a universal approximator of dynamical systems, whose performance often depends on sensitive hyperparameters. Tuning them properly may be difficult and, typically, based on a trial-and-error approach. In this work, we adopt a graph-based framework to interpret and characterize internal dynamics of a class of RNNs called echo state networks (ESNs). We design principled ...

A reconstruction of the last 2,000 years BP of environmental and oceanographic changes on the western margin of Spitsbergen was performed using a multidisciplinary approach including the fossil assemblages of diatoms, planktic and benthic foraminifera and calcareous nannofossils and the use of geochemistry (X-ray fluorescence spectroscopy, X-ray diffraction). We identified two warm periods ...

The main ambition of this thesis was to study the relationship between isotopic signature of fossil foraminifera and past methane emission at Storfjordrenna Pingos site (South of Svalbard) during the last deglaciation in context of past climatic changes. Two gravity cores, the core 1070 from the flank of the presently inactive gas hydrate pingo 5 (GHP5) and the core 1069 from the top of the ...

The performance of piezoelectric sensors deteriorated due to the presence of defect, delamination, and corrosion that needed to be diagnosed for the effective implementation of the structural health monitoring (SHM) framework. A novel experimental approach based on Coulomb coupling is devised to visualise the interaction of ultrasonic waves with microscale defects in the Lead Zirconate Titanate ...

This thesis presents methods for multiscale statistical inference on random fields on a regular two-dimensional lattice. There are two distinct concepts of scale that are used in the thesis. The first one is connected to the computer vision community's understanding of scale-space as a family of smooths of a digital image, with fine structure being revealed at low levels of smoothing and the coarser ...

In this paper, we present the experimental setup and data collection during the Norwegian Radar oil Spill Experiment 2015, followed by a comparison of a subset of the multisensory synthetic aperture radar (SAR) imagery collected during the experiment. Multipolarization SAR data acquired by Radarsat-2, TerraSAR-X, and the uninhabited aerial vehicle synthetic aperture radar (UAVSAR) less than 6 min ...

The mesosphere and lower thermosphere (MLT) region is dominated globally by dynamics at various scales: planetary waves, tides, gravity waves, and stratified turbulence. The latter two can coexist and be significant at horizontal scales less than 500 km, scales that are difficult to measure. This study presents a recently deployed multistatic specular meteor radar system, SIMONe Peru, which can be ...

A statistical model for multilook polarimetric radar data is presented where the polarimetric channels are associated with individual texture variables having potentially different statistical properties. The feasibility of producing closed form probability density functions under certain restrictions is outlined. Mellin kind statistics are derived under various assumptions on the texture variables, ...

Chemometrics was used to develop a multivariate model based on 46 previously reported electrodialytic remediation experiments (EDR) of five different harbour sediments. The model predicted final concentrations of Cd, Cu, Pb and Zn as a function of current density, remediation time, stirring rate, dry/wet sediment, cell set-up as well as sediment properties. Evaluation of the model showed that ...

This work is presenting a fully numerical approach for the direct solution of the Poisson equation for the electrostatic potential given by an arbitrary charge density. Efficient solution of this equation is important in many fields of science, where the current work is dealing with nuclear and electronic potential calculations, used in the field of computational chemistry. The equation is solved ...