Medical Imagery (MI) has made much progress in the lasttwo decades. Making use of new modalities and improvedaccess to technologies, such as magnetic resonance imaging(MRI), has lead to an increasing amount of images andrelated medical data. The increase in data calls for bettercompression methods to support fast communication andstorage of medical images. In MI the most used compression techniques are JPEGand the more modern JPEG-2000 (JP2). Recent attemptsto improve compression has made use of artificial neuralnetworks (ANN) as part of one or more steps in the compressionprocess. This report tries to compress MRI images using a hybridANN approach mostly based on the JP2 approach.The results were validated and compared using both MSEand PSNR as well as compression ratio. A dataset of 500MRI images from one patient was used in training and testingthe implementation. The results of this study were notcomparable to previous work and it fails to even come closeto the JP2 compression rate. This could largely be due toflaws in the implementation or not enough training of theANN, meaning that the proposed method could still be aviable approach for future research.

In this thesis we evaluate four different machine learning algorithms, namely Naive Bayes Classifier, Support Vector Machines, Extreme Learning Machine and Random Forest in the context of stock market investments. The aim is to provide additional information that can be beneficial when creating stock market models to be used in a machine learning setting. All four algorithms are trained on different configurations of data, based on concepts from technical analysis. The configurations contain closing prices, volatility and trading volume in different combinations. These variables are taken from past trading days, where the number of days from which data is to be collected ranges from 2 to 30. The resulting predictors attained from the various algorithms and configurations above reach accuracy rates between $50-54\%$. This thesis concludes that the effect of the different evaluated features vary depending on which algorithm is used as well as how many past trading days are included. Concluding, it is ascertained that the usage of volatility features should at least be considered when building a machine learning model in a stock market context.

This dissertation proposes a combinatorial model, program representations, and constraint solving techniques for integrated register allocation and instruction scheduling in compiler back-ends. In contrast to traditional compilers based on heuristics, the proposed approach generates potentially optimal code by considering all trade-offs between interdependent decisions as a single optimization problem. The combinatorial model is the first to handle a wide array of global register allocation subtasks, including spill code optimization, ultimate coalescing, register packing, and register bank assignment, as well as instruction scheduling for Very Long Instruction Word (VLIW) processors. The model is based on three novel, complementary program representations: Linear Static Single Assignment for global register allocation; copy extension for spilling, basic coalescing, and register bank assignment; and alternative temporaries for spill code optimization and ultimate coalescing. Solving techniques are proposed that exploit the program representation properties for scalability. The model, program representations, and solving techniques are implemented in Unison, a code generator that delivers potentially optimal code while scaling to medium-size functions. Thorough experiments show that Unison: generates faster code (up to 41% with a mean improvement of 7%) than LLVM (a state-of-the-art compiler) for Hexagon (a challenging VLIW processor), generates code that is competitive with LLVM for MIPS32 (a simpler RISC processor), is robust across different benchmarks such as MediaBench and SPECint 2006, scales up to medium-size functions of up to 1000 instructions, and adapts easily to different optimization criteria. The contributions of this dissertation are significant. They lead to a combinatorial approach for integrated register allocation and instruction scheduling that is, for the first time, practical (it robustly scales to medium-size functions) and effective (it yields better code than traditional heuristic approaches). / <p>QC 20141117</p>

A Domain-Specific Language (DSL) is a language tailored for a specific problem domain with the purpose of improving developer productivity and communication with domain experts. In this thesis we investigate how a DSL for normalization of financial derivatives data can be designed and implemented. The thesis includes research on the general subject of DSL engineering and previously established approaches and guidelines for DSL development. We describe a development process that consists of three phases: domain analysis, language design and implementation. The proposed solution was evaluated according to a set of predefined quality criteria. The report concludes with a discussion about data normalization as a DSL domain as well as what impact decisions made during the development process had on the proposed solution. The thesis fulfills its purpose of being an exploratory study of DSL development and the conclusions listed in the final chapter should apply to all data normalization DSLs. / Ett domänspecifikt språk (eng. Domain-Specific Language, DSL) är ett språk som är skräddarsytt för ett specifikt problemområde med syftet att förbättra utvecklares produktivitet och kommunikation med domänexperter. I den här uppsatsen undersöker vi hur ett domänspecifikt språk för normalisering av data som beskriver finansiella derivataffärer kan utformas och implementeras. Uppsatsen omfattar utforskning av det generella ämnet domänspecifika språk och tidigare etablerade tillvägagångssätt och riktlinjer för utveckling av sådana språk. Vi beskriver en utvecklingsproess som består av tre faser: domänanalys, språkutformning och implementation. Den föreslagna lösningen utvärderades enligt en mängd fördefinierade kvalitetskriterier. Rapporten avslutas med en diskussion om datanormalisering som domän för ett domänspecifikt språk och en analys av vilken inverkan beslut som togs under utvecklingsprocessen hade på det slutgiltiga resultatet. Arbetet uppfyller sitt syfte att vara en utforskande studie i DSL-utveckling och slutsatserna som listas i det avslutande kapitlet bör gälla alla domänspecifika språk för normalisering av data.

This thesis presents Bayesian approach for a contextual network anomaly detection. Network anomaly detection is important in a computer system performance monitoring perspective. Detecting a contextual anomaly is much harder since we need to take the context into account in order to explain whether it is normal or abnormal. The main idea of this thesis is to find contextual attributes from a set of indicators, then to estimate the resource loads through the Bayesian model. The proposed algorithm offers three advantages. Firstly, the model can estimate resource loads with automatically selected indicators and its credible intervals. Secondly, both point and collective contextual anomalies can be captured by the posterior predictive distribution. Lastly, the structural interpretation of the model gives us a way to find similar nodes. This thesis employs real data from Radio Network Controller (RNC) to validate the effectiveness in detecting contextual anomalies.

No description available.

The original algorithm proposed by Ford and Fulkerson to solve the maximum flow problem is still in use but is far from the only alternative. This paper introduces that algorithm as well as the similar Edmond-Karp and the more modern Goldberg-Tarjan. Ford-Fulkerson uses depth-first-searches to find augmenting paths through a residual graph. Edmond-Karp instead uses breath-first-searches to achieve a polynomial time complexity. Goldberg-Tarjan solves the problem by gradually pushing flow through the residual graph.  A comparison on randomized graphs where the size, graph density and the maximum edge capacity was varied one at a time was performed. The results show that the Goldberg-Tarjan algorithm has higher performance than the others, but only if it uses heuristics.

Stock trading is increasingly done pseudo-automatically orfully automatically, using algorithms which make day-todayor even moment-to-moment decisions.This report investigates the possibility of creating a virtualstock trader, using a method used in Artificial Intelligence,called Neural Networks, to make intelligent decisions onwhen to buy and sell stocks on the stock market.We found that it might be possible to earn money overa longer period of time, although the profit is less than theaverage stock index. However, the method also performedwell in situations where the stock index is going down. / Aktiehandel genomförs till allt större grad automatiskt ellerhalvautomatiskt, med algoritmer som fattar beslut pådaglig basis eller över ännu kortare tidsintervall.Denna rapport undersöker möjligheten att göra en virtuellaktiehandlare med hjälp av en metod inom artificiellintelligens kallad neurala nätverk, och fatta intelligenta beslutom när aktier på aktiemarknaden ska köpas eller säljas.Vi fann att det är möjligt att tjäna pengar över en längretidsperiod, men vinsten vår algoritm gör över den behandladetidsperioden är mindre än börsindex ökning. Däremotvisar vår algoritm positiva resultat även under sjunkandebörsindex.

This paper studies the performance of height-based weighing functions and compares the results to using the commonly used non height-based weighing functions for holes. For every test performed, the heuristic methods studied in this paper performed better than the commonly used heuristic function. This study also analyses the effect of adding levels of prediction to the heuristic algorithm, which increases the average number of cleared lines by a factor of 85 in total. Utilising these methods can provide increased performance for a Tetris AI. The polynomic weighing functions discussed in this paper provide a performance increase without increasing the needed computation, increasing the number of cleared lines by a factor of 3. The breadth-first search provide a bigger performance increase, but every new level of prediction requires 162 times more computation. Every level increases the number of cleared lines by a factor of 9 from what has been observed in this study.

Recently, much progress has been done towards making vehicles autonomous and to behave in certain ways when interacting with manned and unmanned traffic. Furthermore, autonomous driving is said to be revolutionary and have many benefits. Even though there still are plenty of unsolved issues, several projects have made it possible for groundbreaking steps in the development. This report will cover some of the recently held projects and tools of importance, and describe a conducted simulation of urban driving with autonomous cars. A behavioral steering system and different types of zones is used to implement the simulation in an urban city. The object is to find out how realistic a simulation can be made with models from the gaming industry. With elements such as roundabouts and various crossings, the traffic simulation is evaluated. Some traffic phenomena have been recreated and compared to traffic dynamic models nd real life occurrences. For example, a test ascertains if the simulation can reproduce an expected behavior when stopping at a signalized control. We also try to investigate if the expected performance when driving through a roundabout respectively intersection can be achieved at light and heavy traffic. The results show numerous traffic problems that can behandled by the cars’ behavior system. It is then concluded that it is possible to create a realistic simulation by using models and methods from computer games. / Under den senaste tiden har det gjorts många framsteg inom utvecklingen mot att göra fordon förarlösa. Många verktyghar utvecklats för att få dem bete sig på olika sätt vid interagerandet av bemannad och obemannad trafik. Vidare sägs autonom körning komma med många fördelar, så somfärre trafikolyckor. Flertal projekt och initiativ har gjort det möjligt för banbrytande steg, både i hård- och mjukvara. Dock finns det fortfarande många olösta problem inom dessa områden. Denna rapport kommer att innefatta några av de nyligen utförda projekt som har haft stor betydelse. Ytterligare kommer en genomförd simulering av stadskörning med autonoma bilar att beskrivas. Ett styrsystem för bilarnas beteende, och olika typer av zoner används således för att implementera simuleringen. Målet är att ta reda på hur realistisk en simulering baserad på modeller från spelindustrin kan bli. Med bestånds delar så som rondeller och olika sorters korsningar utvärderas trafiksimuleringen. Några trafikfenomenhar rekonstruerats och jämförts med modeller från trafikdynamiken och verkliga företeelser. Till exempel utrönar ett test om simuleringen kan reproducera ett förväntat beteende vid stoppljus. Vi försöker även se om den förväntade prestandan vid rondellkörning samt körning i korsning uppnås vid lätt och tung trafik. Resultatet visar flertalet trafikproblem som kan hanteras av bilarnas beteendesystem. Slutsatsen kan dras att det ärmöjligt att skapa en realistisk simulering med modeller och metoder från datorspel.