Efficient Optimization Algorithms for Nonlinear Data Analysis

Identification of low-dimensional structures and main sources of variation from multivariate data are fundamental tasks in data analysis. Many methods aimed at these tasks involve solution of an optimization problem. Thus, the objective of this thesis is to develop computationally efficient and theoretically justified methods for solving such problems. Most of the thesis is based on a statistical model, where ridges of the density estimated from the data are considered as relevant features. Finding ridges, that are generalized maxima, necessitates development of advanced optimization methods. An efficient and convergent trust region Newton method for projecting a point onto a ridge of the underlying density is developed for this purpose. The method is utilized in a differential equation-based approach for tracing ridges and computing projection coordinates along them. The density estimation is done nonparametrically by using Gaussian kernels. This allows application of ridge-based methods with only mild assumptions on the underlying structure of the data. The statistical model and the ridge finding methods are adapted to two different applications. The first one is extraction of curvilinear structures from noisy data mixed with background clutter. The second one is a novel nonlinear generalization of principal component analysis (PCA) and its extension to time series data. The methods have a wide range of potential applications, where most of the earlier approaches are inadequate. Examples include identification of faults from seismic data and identification of filaments from cosmological data. Applicability of the nonlinear PCA to climate analysis and reconstruction of periodic patterns from noisy time series data are also demonstrated. Other contributions of the thesis include development of an efficient semidefinite optimization method for embedding graphs into the Euclidean space. The method produces structure-preserving embeddings that maximize interpoint distances. It is primarily developed for dimensionality reduction, but has also potential applications in graph theory and various areas of physics, chemistry and engineering. Asymptotic behaviour of ridges and maxima of Gaussian kernel densities is also investigated when the kernel bandwidth approaches infinity. The results are applied to the nonlinear PCA and to finding significant maxima of such densities, which is a typical problem in visual object tracking.

Viral membrane fusion: is glycoprotein G of rhabdoviruses a representative of a new class of viral fusion proteins?

Enveloped viruses always gain entry into the cytoplasm by fusion of their lipid envelope with a cell membrane. Some enveloped viruses fuse directly with the host cell plasma membrane after virus binding to the cell receptor. Other enveloped viruses enter the cells by the endocytic pathway, and fusion depends on the acidification of the endosomal compartment. In both cases, virus-induced membrane fusion is triggered by conformational changes in viral envelope glycoproteins. Two different classes of viral fusion proteins have been described on the basis of their molecular architecture. Several structural data permitted the elucidation of the mechanisms of membrane fusion mediated by class I and class II fusion proteins. In this article, we review a number of results obtained by our laboratory and by others that suggest that the mechanisms involved in rhabdovirus fusion are different from those used by the two well-studied classes of viral glycoproteins. We focus our discussion on the electrostatic nature of virus binding and interaction with membranes, especially through phosphatidylserine, and on the reversibility of the conformational changes of the rhabdovirus glycoprotein involved in fusion. Taken together, these data suggest the existence of a third class of fusion proteins and support the idea that new insights should emerge from studies of membrane fusion mediated by the G protein of rhabdoviruses. In particular, the elucidation of the three-dimensional structure of the G protein or even of the fusion peptide at different pH's might provide valuable information for understanding the fusion mechanism of this new class of fusion proteins.

Selection of patients for myocardial perfusion scintigraphy based on fuzzy sets theory applied to clinical-epidemiological data and treadmill test results

Coronary artery disease (CAD) is a worldwide leading cause of death. The standard method for evaluating critical partial occlusions is coronary arteriography, a catheterization technique which is invasive, time consuming, and costly. There are noninvasive approaches for the early detection of CAD. The basis for the noninvasive diagnosis of CAD has been laid in a sequential analysis of the risk factors, and the results of the treadmill test and myocardial perfusion scintigraphy (MPS). Many investigators have demonstrated that the diagnostic applications of MPS are appropriate for patients who have an intermediate likelihood of disease. Although this information is useful, it is only partially utilized in clinical practice due to the difficulty to properly classify the patients. Since the seminal work of Lotfi Zadeh, fuzzy logic has been applied in numerous areas. In the present study, we proposed and tested a model to select patients for MPS based on fuzzy sets theory. A group of 1053 patients was used to develop the model and another group of 1045 patients was used to test it. Receiver operating characteristic curves were used to compare the performance of the fuzzy model against expert physician opinions, and showed that the performance of the fuzzy model was equal or superior to that of the physicians. Therefore, we conclude that the fuzzy model could be a useful tool to assist the general practitioner in the selection of patients for MPS.

Local Binary Patterns in Focal-Plane Processing. Analysis and Applications

Feature extraction is the part of pattern recognition, where the sensor data is transformed into a more suitable form for the machine to interpret. The purpose of this step is also to reduce the amount of information passed to the next stages of the system, and to preserve the essential information in the view of discriminating the data into different classes. For instance, in the case of image analysis the actual image intensities are vulnerable to various environmental effects, such as lighting changes and the feature extraction can be used as means for detecting features, which are invariant to certain types of illumination changes. Finally, classification tries to make decisions based on the previously transformed data. The main focus of this thesis is on developing new methods for the embedded feature extraction based on local non-parametric image descriptors. Also, feature analysis is carried out for the selected image features. Low-level Local Binary Pattern (LBP) based features are in a main role in the analysis. In the embedded domain, the pattern recognition system must usually meet strict performance constraints, such as high speed, compact size and low power consumption. The characteristics of the final system can be seen as a trade-off between these metrics, which is largely affected by the decisions made during the implementation phase. The implementation alternatives of the LBP based feature extraction are explored in the embedded domain in the context of focal-plane vision processors. In particular, the thesis demonstrates the LBP extraction with MIPA4k massively parallel focal-plane processor IC. Also higher level processing is incorporated to this framework, by means of a framework for implementing a single chip face recognition system. Furthermore, a new method for determining optical flow based on LBPs, designed in particular to the embedded domain is presented. Inspired by some of the principles observed through the feature analysis of the Local Binary Patterns, an extension to the well known non-parametric rank transform is proposed, and its performance is evaluated in face recognition experiments with a standard dataset. Finally, an a priori model where the LBPs are seen as combinations of n-tuples is also presented

Scalable Feature Selection Applications for Genome-Wide Association Studies of Complex Diseases

Personalized medicine will revolutionize our capabilities to combat disease. Working toward this goal, a fundamental task is the deciphering of geneticvariants that are predictive of complex diseases. Modern studies, in the formof genome-wide association studies (GWAS) have afforded researchers with the opportunity to reveal new genotype-phenotype relationships through the extensive scanning of genetic variants. These studies typically contain over half a million genetic features for thousands of individuals. Examining this with methods other than univariate statistics is a challenging task requiring advanced algorithms that are scalable to the genome-wide level. In the future, next-generation sequencing studies (NGS) will contain an even larger number of common and rare variants. Machine learning-based feature selection algorithms have been shown to have the ability to effectively create predictive models for various genotype-phenotype relationships. This work explores the problem of selecting genetic variant subsets that are the most predictive of complex disease phenotypes through various feature selection methodologies, including filter, wrapper and embedded algorithms. The examined machine learning algorithms were demonstrated to not only be effective at predicting the disease phenotypes, but also doing so efficiently through the use of computational shortcuts. While much of the work was able to be run on high-end desktops, some work was further extended so that it could be implemented on parallel computers helping to assure that they will also scale to the NGS data sets. Further, these studies analyzed the relationships between various feature selection methods and demonstrated the need for careful testing when selecting an algorithm. It was shown that there is no universally optimal algorithm for variant selection in GWAS, but rather methodologies need to be selected based on the desired outcome, such as the number of features to be included in the prediction model. It was also demonstrated that without proper model validation, for example using nested cross-validation, the models can result in overly-optimistic prediction accuracies and decreased generalization ability. It is through the implementation and application of machine learning methods that one can extract predictive genotype–phenotype relationships and biological insights from genetic data sets.

Organization of the marketing communication activities after fusion - the integration and the decision making processes

This master’s thesis was made in order to gain answers to the question of how the integration of the marketing communications and the decision making related to it in a geographically dispersed service organization could be improved in a situation where an organization has gone through a merger. The effects of the organizational design dimensions towards the integration of the marketing communications and the decision making related to it was the main focus. A case study as a research strategy offered a perfect frames for an exploratory study and the data collection was conducted by semi-structured interviews and observing. The main finding proved that from the chosen design dimensions, decentralization, coordination and power, could be found specific factors that in a geographically dispersed organization are affecting the integration of the marketing communications negatively. The effects can be seen mostly in the decision making processes, roles and in the division of responsibility, which are affecting the other dimensions and by this, the integration. In a post-merger situation, the coordination dimension and especially the information asymmetry and the information flow seem to have a largest affect towards the integration of the marketing communications. An asymmetric information distribution with the lack of business and marketing education resulted in low self-assurance and at the end in fragmented management and to the inability to set targets and make independent decisions. As conclusions it can be stated, that with the organizational design dimensions can the effects of a merger towards the integration process of the marketing communications to be evaluated.

Characterization of Brazilian wheat cultivars for specific technological applications

Functional and technological properties of wheat depend on its chemical composition, which together with structural and microscopic characteristics, define flour quality. The aim of the present study was to characterize four Brazilian wheat cultivars (BRS Louro, BRS Timbauva, BRS Guamirim and BRS Pardela) and their respective flours in order to indicate specific technological applications. Kernels were analyzed for test weight, thousand kernel weight, hardness, moisture, and water activity. Flours were analyzed for water activity, color, centesimal composition, total dietary fiber, amylose content and identification of high molecular weight glutenins. The rheological properties of the flours were estimated by farinography, extensography, falling number, rapid visco amylography, and glutomatic and glutork equipment. Baking tests and scanning electron microscopy were also performed. The data were subjected to analysis of variance and principal component analysis. BRS Timbauva and BRS Guamirim presented results that did not allow for specific technological application. On the other hand, BRS Louro presented suitable characteristics for the elaboration of products with low dough strength such as cakes, pies and biscuits, while BRS Pardela seemed suitable for bread and pasta products.

Effect of different stress conditions on the stability of quercetin-loaded lipid microparticles produced with babacu ( Orbignya speciosa ) oil: evaluation of their potential use in food applications

Lipid micro and nanoparticles have been extensively investigated as carriers for hydrophobic bioactives in food systems because they can simultaneously increase the dispersibility of these lipophilic substances and help improve their bioavailability. In this study, lipid microparticles of babacu oil and denatured whey protein isolate were produced, and their ability to protect quercetin against degradation was evaluated over 30 days of storage. Additionally, the lipid microparticles were subjected to the typical stress conditions of food processing (presence of sucrose, salt, and thermal stresses), and their physico-chemical stability was monitored. The data show that the babacu microparticles efficiently avoided the oxidation of quercetin because 85% of the initial amount of the flavonoid was preserved after 30 days. The particles were notably stable up to a temperature of 70 °C for 10 minutes at relatively high concentrations of salt and sucrose. The type of stirring (mechanical or magnetic) also strongly affected the stability of the dispersions.

A systematic mapping study on open data

This thesis presented the overview of Open Data research area, quantity of evidence and establishes the research evidence based on the Systematic Mapping Study (SMS). There are 621 such publications were identified published between years 2005 and 2014, but only 243 were selected in the review process. This thesis highlights the implications of Open Data principals’ proliferation in the emerging era of the accessibility, reusability and sustainability of data transparency. The findings of mapping study are described in quantitative and qualitative measurement based on the organization affiliation, countries, year of publications, research method, star rating and units of analysis identified. Furthermore, units of analysis were categorized by development lifecycle, linked open data, type of data, technical platforms, organizations, ontology and semantic, adoption and awareness, intermediaries, security and privacy and supply of data which are important component to provide a quality open data applications and services. The results of the mapping study help the organizations (such as academia, government and industries), re-searchers and software developers to understand the existing trend of open data, latest research development and the demand of future research. In addition, the proposed conceptual framework of Open Data research can be adopted and expanded to strengthen and improved current open data applications.

Adopting the Vaadin Java framework to rebuild VTT’s Web Interfaces and Database Applications

With the growth in new technologies, using online tools have become an everyday lifestyle. It has a greater impact on researchers as the data obtained from various experiments needs to be analyzed and knowledge of programming has become mandatory even for pure biologists. Hence, VTT came up with a new tool, R Executables (REX) which is a web application designed to provide a graphical interface for biological data functions like Image analysis, Gene expression data analysis, plotting, disease and control studies etc., which employs R functions to provide results. REX provides a user interactive application for the biologists to directly enter the values and run the required analysis with a single click. The program processes the given data in the background and prints results rapidly. Due to growth of data and load on server, the interface has gained problems concerning time consumption, poor GUI, data storage issues, security, minimal user interactive experience and crashes with large amount of data. This thesis handles the methods by which these problems were resolved and made REX a better application for the future. The old REX was developed using Python Django and now, a new programming language, Vaadin has been implemented. Vaadin is a Java framework for developing web applications and the programming language is extremely similar to Java with new rich components. Vaadin provides better security, better speed, good and interactive interface. In this thesis, subset functionalities of REX was selected which includes IST bulk plotting and image segmentation and implemented those using Vaadin. A code of 662 lines was programmed by me which included Vaadin as the front-end handler while R language was used for back-end data retrieval, computing and plotting. The application is optimized to allow further functionalities to be migrated with ease from old REX. Future development is focused on including Hight throughput screening functions along with gene expression database handling

Use of airborne laser scanner data in demanding forest conditions

Most of the applications of airborne laser scanner data to forestry require that the point cloud be normalized, i.e., each point represents height from the ground instead of elevation. To normalize the point cloud, a digital terrain model (DTM), which is derived from the ground returns in the point cloud, is employed. Unfortunately, extracting accurate DTMs from airborne laser scanner data is a challenging task, especially in tropical forests where the canopy is normally very thick (partially closed), leading to a situation in which only a limited number of laser pulses reach the ground. Therefore, robust algorithms for extracting accurate DTMs in low-ground-point-densitysituations are needed in order to realize the full potential of airborne laser scanner data to forestry. The objective of this thesis is to develop algorithms for processing airborne laser scanner data in order to: (1) extract DTMs in demanding forest conditions (complex terrain and low number of ground points) for applications in forestry; (2) estimate canopy base height (CBH) for forest fire behavior modeling; and (3) assess the robustness of LiDAR-based high-resolution biomass estimation models against different field plot designs. Here, the aim is to find out if field plot data gathered by professional foresters can be combined with field plot data gathered by professionally trained community foresters and used in LiDAR-based high-resolution biomass estimation modeling without affecting prediction performance. The question of interest in this case is whether or not the local forest communities can achieve the level technical proficiency required for accurate forest monitoring. The algorithms for extracting DTMs from LiDAR point clouds presented in this thesis address the challenges of extracting DTMs in low-ground-point situations and in complex terrain while the algorithm for CBH estimation addresses the challenge of variations in the distribution of points in the LiDAR point cloud caused by things like variations in tree species and season of data acquisition. These algorithms are adaptive (with respect to point cloud characteristics) and exhibit a high degree of tolerance to variations in the density and distribution of points in the LiDAR point cloud. Results of comparison with existing DTM extraction algorithms showed that DTM extraction algorithms proposed in this thesis performed better with respect to accuracy of estimating tree heights from airborne laser scanner data. On the other hand, the proposed DTM extraction algorithms, being mostly based on trend surface interpolation, can not retain small artifacts in the terrain (e.g., bumps, small hills and depressions). Therefore, the DTMs generated by these algorithms are only suitable for forestry applications where the primary objective is to estimate tree heights from normalized airborne laser scanner data. On the other hand, the algorithm for estimating CBH proposed in this thesis is based on the idea of moving voxel in which gaps (openings in the canopy) which act as fuel breaks are located and their height is estimated. Test results showed a slight improvement in CBH estimation accuracy over existing CBH estimation methods which are based on height percentiles in the airborne laser scanner data. However, being based on the idea of moving voxel, this algorithm has one main advantage over existing CBH estimation methods in the context of forest fire modeling: it has great potential in providing information about vertical fuel continuity. This information can be used to create vertical fuel continuity maps which can provide more realistic information on the risk of crown fires compared to CBH.

Possibilities and challenges in utilization of information provided by Industrial Internet applications

Global digitalization has affected also industrial sector. A trend called Industrial Internet has been present for some years and established relatively steady position in businesses. Industrial Internet is also referred with the terminology Industry 4.0 and in consumer businesses IoT (Internet of Things). Eventually, trend consists of many traditionally proven technologies and concepts, such as condition monitoring, remote services, predictive maintenance and Internet customer portals. All these technologies and information related to them are estimated to change the rules of business in industrial sector. This may result even a new industrial revolution. This research has its focus on Industrial Internet products, services and applications. The study analyses four case companies and their digital service offerings. According to this analysis the comparison of these services is done to find out if there is still space for companies to gain competitive advantage through differentiation with these state of the art solutions. One of the case companies, Case Company Ltd., is working as a primary case company and a subscriber of this particular research. The research and results are analyzed primarily from this company’s perspective and need. In empirical part, the research clarifies how Case Company Ltd. has allocated its development resources through last five years. These allocations in certain categories are then compared to other case companies’ current customer offering and conclusions are made how the approach of different companies differ from each other. Existing theoretical knowledge of Industrial Internet is about to find its shape. In this research we take a look how the case company analysis and findings correlate with the existing knowledge and literature of the topic.

Quadtree representation and compression of spatial data

Spatial data representation and compression has become a focus issue in computer graphics and image processing applications. Quadtrees, as one of hierarchical data structures, basing on the principle of recursive decomposition of space, always offer a compact and efficient representation of an image. For a given image, the choice of quadtree root node plays an important role in its quadtree representation and final data compression. The goal of this thesis is to present a heuristic algorithm for finding a root node of a region quadtree, which is able to reduce the number of leaf nodes when compared with the standard quadtree decomposition. The empirical results indicate that, this proposed algorithm has quadtree representation and data compression improvement when in comparison with the traditional method.

Towards Evaluating and Modelling the Impacts of Mobile-Based Augmented Reality Applications on Learning and Engagement

Mobile augmented reality applications are increasingly utilized as a medium for enhancing learning and engagement in history education. Although these digital devices facilitate learning through immersive and appealing experiences, their design should be driven by theories of learning and instruction. We provide an overview of an evidence-based approach to optimize the development of mobile augmented reality applications that teaches students about history. Our research aims to evaluate and model the impacts of design parameters towards learning and engagement. The research program is interdisciplinary in that we apply techniques derived from design-based experiments and educational data mining. We outline the methodological and analytical techniques as well as discuss the implications of the anticipated findings.

Towards Evaluating and Modelling the Impacts of Mobile-Based Augmented Reality Applications on Learning and Engagement

Mobile augmented reality applications are increasingly utilized as a medium for enhancing learning and engagement in history education. Although these digital devices facilitate learning through immersive and appealing experiences, their design should be driven by theories of learning and instruction. We provide an overview of an evidence-based approach to optimize the development of mobile augmented reality applications that teaches students about history. Our research aims to evaluate and model the impacts of design parameters towards learning and engagement. The research program is interdisciplinary in that we apply techniques derived from design-based experiments and educational data mining. We outline the methodological and analytical techniques as well as discuss the implications of the anticipated findings.