Luftstridsskolan upplever problem med deras digitala resurser eftersom de saknar överblick och vet inte hur studerande använder dessa. Infor- mationsarkitektur är ett område som bemöter dessa problem och dess metodiska ramverk kommer i detta arbete att användas för att kartlägga resurser, se hur dessa används och slutligen skapa en ny struktur att pre- sentera resurserna i.

This thesis develops a Gaussian processes model for bankruptcy risk classification and prediction in a Bayesian framework. Gaussian processes and linear logistic models are discriminative methods used for classification and prediction purposes. The Gaussian processes model is a much more flexible model than the linear logistic model with smoothness encoded in the kernel with the potential to improve the modeling of the highly nonlinear relationships between accounting ratios and bankruptcy risk. We compare the linear logistic regression with the Gaussian process classification model in the context of bankruptcy prediction. The posterior distributions of the GPs are non-Gaussian, and we investigate the effectiveness of the Laplace approximation and the expectation propagation approximation across several different kernels for the Gaussian process. The approximate methods are compared to the gold standard of Markov Chain Monte Carlo (MCMC) sampling from the posterior. The dataset is an unbalanced panel consisting of 21846 yearly observations for about 2000 corporate firms in Sweden recorded between 1991−2008. We used 5000 observations to train the models and the rest for evaluating the predictions. We find that the choice of covariance kernel affects the GP model’s performance and we find support for the squared exponential covariance function (SEXP) as an optimal kernel. The empirical evidence suggests that a multivariate Gaussian processes classifier with squared exponential kernel can effectively improve bankruptcy risk prediction with high accuracy (90.19 percent) compared to the linear logistic model (83.25 percent).

The purpose of the project was to make a system that could extract data from a mobile eye tracker and make it comparable with data from visual stimuli. The produced system was programmed in Java and provided all the necessary parts that were required to achieve the purpose. This provides a foundation for further research to determine whether the eye tracker is sufficiently accurate to diagnose Parkinson’s disease.

Automation of model building enables new predictive models to be generated in a faster, easier and more straightforward way once new data is available to predict on. Automation can also reduce the demand for tedious bookkeeping that is generally needed in manual workflows (e.g. intermediate files needed to be passed between steps in a workflow). The applicability of the Spark framework related to the creation of pipelines for predictive drug discovery was here evaluated and resulted in the implementation of two pipelines that serves as a proof of concept. Spark is considered to provide good means of creating pipelines for pharmaceutical purposes and its high level approach to distributed computing reduces the effort put on the developer compared to a regular HPC implementation.

Parallel computing is on the rise and the applications are steadily growing. This thesis will consider one such application, namely computational electromagnetic modeling (CEM). The methods that are being used as of today are usually computationally heavy which makes them time-consuming and not considered as viable options for some applications. In this thesis a parallel solution of the FDTD method has been presented. In the results it is shown to be faster than the sequential solutions upon which it is based, and alterations are suggested which could improve it further.

During the last two decades the interest about computer vision raised steadily with multiple applications in fields like medical care, automotive, entertainment, retail, industrial, and security. Objectdetection is part of the recognition problem, which is the most important scope of the computervision environment.The target of this thesis work is to analyse and propose a solution for object detection in a real timedynamic environment. RoboCup@Home will be the benchmarking event for this system, which willbe equipped on a robot competing in the 2018 event. The system has to be robust and fast enoughto allow the robot to react to each environment change in a reasonable amount of time.The input hardware used to achieve such system comprise of a Microsoft Kinect, which providesan high definition camera and fast and reliable 3D scanner. Through the study and analysis ofstate-of-the-art algorithms regarding machine vision and object recognition, the more suitable oneshave been tested to optimise the execution on the targeted hardware. Porting of the application toan embedded platform is discussed.

Fly or Die was a first year project that became fairly successful. It won the Gamer’s Choice Award at the Swedish Game Awards 2013 and has been on display in many locations, both in Sweden and abroad. Recreating the game and porting it from SFML to Unity was sparked by winning the award and by all the attention that the game received prior to the conversion pro-cess. The final version requires the Unity 5 Personal Edition and Microsoft Visual Studio 2013. The conversion process uses the old code base as a template and attempts to incorporate the same coding principles. In the new versions, all levels are built using many of the 2D fea-tures that Unity 5 offers, including that the game can be played with two to four players, and that some of the levels are enhanced or changed by adding additional graphics or game play enhancements. An approach to how the levels have been created and a comparison between the new and the old version is presented, including the differences, and why it was necessary to change certain aspects of some of the levels. The author thus recommends porting a game from a previous platform to Unity 5 in order to make it playable on a larger number of plat-forms, since the advantages of using an engine like Unity far outweigh the disadvantages.

In this thesis digital image analysis is applied to Scanning Electron Microscope imagesof dispersion barriers to measure specific properties. The thin barriers are used asprotection for paperboard packaging and are made of polymers and fillers. The orientation, area, length and density distributions of the fillers determine the functionality and quality of the barrier. Methods built on image analysis tools are developed with the objective to measure these quantities. Input for the methods are Scanning Electron Microscope images showing the cross-section of the barriers. To make the images relevant for the methods they are preprocessed by reducing noise and distinguishing fillers from the background. For measuring the orientation distribution of the fillers two different methods are implemented and compared. The first one is based on a structure tensor and the other one applies a covariance matrix. The structure tensor is preferable because of its flexibility and better performance for complex images. The area and length distributions are measured by applying mathematical morphology together withsoft-clipping. The density distribution is obtained by filtering the underlying image twice with a uniform filter which creates a heat map. The developed methods are evaluated by applying them on fabricated binary test images with known properties. The methods are very accurate when applied on simple test images but for more complex test images with greater variation the accuracy decreases. However, for most applications the results are still on an acceptable level.

This thesis examines ways to improve the user experience of a mobile Internetconnection. We discuss how this can be achieved using dierent techniques,but primarily focus on techniques using network performance maps. We designand implement a proof-of-concept system that allows users to record networkperformance experience and share these records among each other such as toimprove their future performance.We use bandwidth measurements at locations over an area to create a networkperformance map. We examine what such maps can be used for and howit is best implemented. We also create a proof-of-concept implementation fora system creating network performance maps using multiple mobile devices.The thesis also discusses how information sharing between mobile devices usingWi-Fi networks can be done eciently. This is also something we use inour proof-of-concept implementation, when synchronizing data between mobiledevices.In summary, our Android-based proof-of-concept implementation allows usersto perform bandwidth measurements, and to share these measurements amongeach other, such as to help each other improve their network performance maps.A good such system allows users to get an overview of the network performanceover the measured area; information that can be used to optimize the Internetusage via download scheduling or bit rate adaption.

Den här rapporten, som är en del av kursen TDDD83 Kandidatprojekt datateknik vid Linköpings Tekniska Högskola, beskriver utvecklandet av ett e-handelssystem för den fiktiva t-shirtbutiken Utique. Den berör dels tekniska aspekter, men även de erfarenheter och lärdomar de tre gruppmedlemmarna fått under projektets gång. För att genomföra projektet användes programmeringsspråken Html, Css, Php, Mysql och Javascript samt ramverken och biblioteken Bootstrap, Cakephp och Jquery. Under projektets gång användes, efter anpassning till gruppstorleken, den agila utvecklingsmetoden Scrum, vilket beskrivs kortfattat i rapporten. Rapporten innefattar även en marknadsföringsplan för Utique där en konkurrensanalys genomförts och ämnena marknadssegmentering, positionering och differentiering berörts. Marknadsföringsplanen avslutas med en marknadsmix. Avslutningsvis, diskuteras utvecklingsområden för systemet, styrkor och svagheter med koden samt svårigheter som dykt upp under projektets gång.