We compared the performance of Sentinel-2 and Landsat 8 data for forest variable prediction in boreal Southern Finland. defined twelve modelling setups to train multivariable models with either multilayer perceptron (MLP) or regression tree brute force forward selection method. The reference consisted 739 circular field plots that had been collected by Finnish Forest Centre concurrently acquisitions. input were divided into training, validation test sets equal sizes 100 iterations each...

Traditional design of experiments and response surface methodology are widely used in engineering process development. Bayesian optimization is an alternative machine learning approach that adaptively selects successive experimental conditions based on a predefined performance measure. Here we compared the two approaches using simulations empirical alkaline wood delignification to identify important benefits drawbacks context experiments. The showed selection initial measurement noise...

We search for methods or tools to detect whether the 1-dimensional marginal distribution of traffic increments aggregate TCP-traffic satisfy hypothesis approximate normality. Gaussian approximation requires a high level aggregation in both "vertical" (source aggregation) and "horizontal" (time scale) directions. discuss these different concepts first separately, with an example from real data traffic, show how rule out cases where will not be sufficient. is then quantified square linear...

Performance of the above ground (dry) biomass estimation with medium resolution optical (ALOS AVNIR) data and radar PALSAR) was evaluated on a tropical forest site in Lao PDR (Laos). The average reference plots relatively low, 78 t/ha, due to strong anthropogenic influence most study area. estimates were computed using linear regression analysis Probability method that combines unsupervised clustering fuzzy estimation. predictions validated independent field plot data. With all methods...

This paper describes an improved concept for the mapping of tropical forest classes with ALOS AVNIR and PALSAR data. The improvement comes from a combination sample very high resolution (VHR) satellite images medium wall-to-wall in statistical sampling framework. approach developed makes it possible to obtain reliable information on accuracy over whole area interest. A simulation study indicated that VHR should be collected stratified manner using small (25 km) images. cover approximately...

The applicability of optical and synthetic aperture radar (SAR) data for land cover classification to support REDD+ (Reducing Emissions from Deforestation Forest Degradation) MRV (measuring, reporting verification) services was tested on a tropical sub-tropical test site. 100 km by site situated in the State Chiapas Mexico. Land classifications were computed using RapidEye Landsat TM satellite images ALOS PALSAR L-band Envisat ASAR C-band images. Identical sample plot Kompsat-2 imagery...

Static Wireless Mesh Networks (WMN) based on high capacity pencil-beam millimeter wave (mmWave) links offer an attractive solution for small cell backhauling in 5G networks. However, the delay targets are such that it is quite challenging to develop adequate link scheduling and routing mechanisms. Furthermore, susceptibility of mmWave radio adverse weather conditions other obstructions line-of-sight line makes necessary able cope with frequent degradations failures. We propose approach...

Spatial-spectral approaches applied on hyperspectral images (HSI) with limited labels suffer from overfitting when the size of input filters and percentage training data increases. In those cases, pixel values corresponding to testing sets are partly or completely seen during phase, reducing number independent pixels leading overoptimistic accuracy assessment. These effects have been demonstrated in several previous works but still require attention. this work we propose additional...

A method to detect abrupt land cover changes using hierarchical clustering of multi-temporal satellite imagery was developed. The Autochange outputs the pre-change class, change magnitude, and type. Pre-change information is transferred post-change based on classes derived by unsupervised clustering, enabling data from different instruments for pre- post-change. magnitude types are computed image within each cluster, comparing mean intensity values lower level clusters with their parent...

A new link scheduling approach for millimeter wave wireless mesh network is presented. The proposed solution designed to work together with a multiple overlapping spanning trees-based routing schemes on topology. schedule computation will minimize the end-to-end delay along subset of shortest-hop paths from non-gateway nodes gateway node mesh, where each path obtained one trees. itself consists cyclic repetition vector transmission sets, set maximal collection links that can be active...

This paper explores network design requirements and radio resource management (RRM) functionalities required to enable a heterogeneous millimeter-wave (mmW) in both backhaul access links. First, the envisioned of mmW is outlined highlighting main functional blocks key differentiating factors from non-mmW architectures. Next, RRM networks are discussed detail, focusing on routing, link scheduling, dynamic TDD "beam cell" Finally, some recent findings presented, impact identified network-wide...

Abstract We study the problem of making algorithmic statistical inferences about dynamics city traffic. Our data is based on loop detector counts observed vehicles in various roads Tampere, Finland. show that meaningful correlations can be found between traffic asymmetries at different measurement locations. The asymmetry difference opposite directions a road. further quantified by estimating how much they effect average values neighbouring Conditional expectations, both sample and binormal...

Abstract Spectral scatterometry is a technique that allows rapid measurements of diffraction efficiencies diffractive optical elements (DOEs). The analysis such has traditionally been laborious and time consuming. However, machine learning can be employed to aid in the measured efficiencies. In this paper we describe novel system for providing multiple measurands rapidly concurrently using spectral scatterometer an artificial neural network (ANN) which trained utilising transfer learning....

Our objective was to develop a method for the assessment of forest area and structural variables cases in which availability representative ground reference data is poor these are not collected from whole interest. We implemented two independent approaches estimation European boreal forest: (i) computation wall-to-wall estimates using moderate- low-resolution VIIRS imagery Suomi NPP mission; (ii) visual interpretation plots samples very high resolution (VHR) satellite obtained via two-stage...

Abstract A doubly stochastic, conditionally binomial model is proposed to describe volumes of vehicular origin-destination flows in regular traffic, such as morning rush hours. The statistical properties this are motivated by the data obtained from inductive loop traffic counts. parameters can be expressed rational functions first and second order moments observed link Challenges arising inaccuracy moment estimates studied. real problem Tampere city solved optimisation methods accuracy...

Supervising quality of service will become increasingly important as critical applications switch to use wireless communication services such 5G URLLC. This is also crucial for automation ports, with focus here on automated handling containers in enabling efficient and safe marine logistics. In order move from fixed connections wireless, high performance testing methods are needed. We present a framework passive QoS measurements measurement setup measure delay loss. estimate the empirical...