Gaussian graphical model has been a popular tool to investigate conditional dependency between random variables by estimating sparse precision matrices. Two problems have been discussed. One is to learn multiple Gaussian graphical models at multilevel from unknown classes. Another one is to select Gaussian process in semiparametric multi-kernel machine regression. The first problem is approached by Gaussian graphical model. In this project, I consider learning multiple connected graphs among multilevel variables from unknown classes. I esti- mate the classes of the observations from the mixture distributions by evaluating the Bayes factor and learn the network structures by fitting a novel neighborhood selection algorithm. This approach is able to identify the class membership and to reveal network structures for multilevel variables simultaneously. Unlike most existing methods that solve this problem by frequentist approaches, I assess an alternative to a novel hierarchical Bayesian approach to incorporate prior knowledge. The second problem focuses on the analysis of correlated high-dimensional data which has been useful in many applications. In this work, I consider a problem of detecting signals with a semiparametric regression model which can study the effects of fixed covariates (e.g. clinical variables) and sets of elements (e.g. pathways of genes). I model the unknown high-dimension functions of multi-sets via multi-Gaussian kernel machines to consider the possibility that elements within the same set interact with each other. Hence, my variable selection can be considered as Gaussian process selection. I develop my Gaussian process selection under the Bayesian variable selection framework. / Doctor of Philosophy / A network can be represented by nodes and edges between nodes. Under the assumption of multivariate Gaussian distribution, a graphical model is called a Gaussian graphical model, where edges are undirected. Gaussian graphical model has been studied for years to understand conditional dependency structure between random variables. Two problems have been discussed. In the first project, I consider learning multiple connected graphs among multilevel variables from unknown classes. I estimate the classes of the observations from the mixture distributions. This approach is able to identify the class membership and to reveal network structures for multilevel variables simultaneously. Unlike most existing methods that solve this problem by frequentist approaches, I assess an alternative to a novel hierarchical Bayesian approach to incorporate prior knowledge. The second problem focuses on the analysis of correlated high-dimensional data which has been useful in many applications. In this work, I consider a problem of detecting signals with a semiparametric regression model which can study the effects of fixed covariates (e.g. clinical variables) and sets of elements (e.g. pathways of genes). I model the unknown high-dimension functions of multi-sets via multi-Gaussian kernel machines to consider the possibility that elements within the same set interact with each other. Hence, my variable selection can be considered as Gaussian process selection. I develop my Gaussian process selection under the Bayesian variable selection framework

We study maximum entropy correlated equilibria in (multi-player)games and provide two gradient-based algorithms that are guaranteedto converge to such equilibria. Although we do not provideconvergence rates for these algorithms, they do have strong connectionsto other algorithms (such as iterative scaling) which are effectiveheuristics for tasks such as statistical estimation.

No description available.

Development of advanced techniques for biological network visualisation is crucial for successful progress in the areas of systems-level biology and data-intensive bioinformatics. However, current techniques for biological network visualisation fall short of expectations for representing extensive biological networks. In order to provide really useful network visualisation tools, new approaches have to be proposed and applied alongside with those most powerful features of current visualisation systems. The resulting representation techniques have to be tested by applying to large-scale examples that would include metabolic, signaling and gene expression events. User survey should also be carried out to further prove the advantages of the new techniques. The present thesis describes an attempt to achieve the above objectives, by performing the following steps: 1) existing approaches in the area of network representation were analyzed and their shortcomings and advantages were defined; 2) new notation has been developed, in which, the defined best features of the existing systems were integrated with newly introduced potent features such as compact visualization, ‘functional gate’ and ‘identity gate’, 4) new framework was developed that allows managing large-scale networks that are represented on different levels of details and different levels of constrains, while keeping each diagram semantically unambiguous, 5) extensive examples, including genome-scaled human metabolic network and TNF-alpha receptor signalling network, were used to prove that the designed notation and the framework can be applied efficiently, and, finally, 6) a notation survey has been carried out to validate the advantages of the newly developed notation over the existing ones.

Motivation: Next-generation sequencing technology is increasingly being used for clinical diagnostic tests. Unlike research cell lines, clinical samples are often genomically heterogeneous due to low sample purity or the presence of genetic subpopulations. Therefore, a variant calling algorithm for calling low-frequency polymorphisms in heterogeneous samples is needed. Result: We present a novel variant calling algorithm that uses a hierarchical Bayesian model to estimate allele frequency and call variants in heterogeneous samples. We show that our algorithm improves upon current classifiers and has higher sensitivity and specificity over a wide range of median read depth and minor allele frequency. We apply our model and identify twelve mutations in the PAXP1 gene in a matched clinical breast ductal carcinoma tumor sample; two of which are loss-of-heterozygosity events.

This Master’s thesis presents an algorithm that finds the semantic differences between two versions of files, an older version and a new modified version of the file. The algorithm is responsible for finding changes in the program’s behavior and displaying them graphically. By this a lot of time can be saved because it is not necessary to go through the whole file to find the differences.   The program, Semantic Diff, developed in this master thesis uses the Javacc parser generator which is used to parse files and generate the abstract syntax tree for them. Using this tree it is possible to see all the methods, classes, constructors and parameters for both older version and modified version.  By comparing all the methods, classes and interfaces of both the versions it is possible to find the differences that change the program behavior.   The algorithm for finding semantic differences has been evaluated by testing it on various test cases. By making changes in the original file and in the modified file. Like adding methods and deleting methods and adding classes to the files. The algorithm highlights those methods with green color which are added newly in the modified file and highlights the methods with red color in the original file which got deleted in modified file. This algorithm also finds the textual difference between two files and highlights those lines which are changed in modified file and which got deleted from original file.

When it comes to simulator training, there are two main target groups, the student and the teacher/instructor. This thesis will cover usability aspects for both of these groups when it comes to tasks related to simulator training. For the student it will mainly be about how they can interact with the interface from the simulator and how results and feedback from exercises are presented to them. For the teacher/instructor on the other hand, usability aspects for managing the students and the exercises will be addressed, along with how results and progress shall be presented in a way that is easy to grasp and understand. A redesign of the current system used to manage these issues will be preformed. Some of the aspects that the redesign will cover are usability aspects, graphical design and workflow. One of the main outcomes from this thesis is a system that supports many different levels of engagement from the user, allowing users with different background and interest to interact with the system as effortless as possible/wanted.

Building intelligent systems that are capable of extracting high-level representations from high-dimensional sensory data lies at the core of solving many AI related tasks, including object recognition, speech perception, and language understanding. Theoretical and biological arguments strongly suggest that building such systems requires models with deep architectures that involve many layers of nonlinear processing. The aim of the thesis is to demonstrate that deep generative models that contain many layers of latent variables and millions of parameters can be learned efficiently, and that the learned high-level feature representations can be successfully applied in a wide spectrum of application domains, including visual object recognition, information retrieval, and classification and regression tasks. In addition, similar methods can be used for nonlinear dimensionality reduction.

Building intelligent systems that are capable of extracting high-level representations from high-dimensional sensory data lies at the core of solving many AI related tasks, including object recognition, speech perception, and language understanding. Theoretical and biological arguments strongly suggest that building such systems requires models with deep architectures that involve many layers of nonlinear processing. The aim of the thesis is to demonstrate that deep generative models that contain many layers of latent variables and millions of parameters can be learned efficiently, and that the learned high-level feature representations can be successfully applied in a wide spectrum of application domains, including visual object recognition, information retrieval, and classification and regression tasks. In addition, similar methods can be used for nonlinear dimensionality reduction.

<p>This Master’s thesis presents an algorithm that finds the semantic differences between two versions of files, an older version and a new modified version of the file. The algorithm is responsible for finding changes in the program’s behavior and displaying them graphically. By this a lot of time can be saved because it is not necessary to go through the whole file to find the differences.</p><p> </p><p>The program, Semantic Diff, developed in this master thesis uses the Javacc parser generator which is used to parse files and generate the abstract syntax tree for them. Using this tree it is possible to see all the methods, classes, constructors and parameters for both older version and modified version.  By comparing all the methods, classes and interfaces of both the versions it is possible to find the differences that change the program behavior.</p><p> </p><p>The algorithm for finding semantic differences has been evaluated by testing it on various test cases. By making changes in the original file and in the modified file. Like adding methods and deleting methods and adding classes to the files. The algorithm highlights those methods with green color which are added newly in the modified file and highlights the methods with red color in the original file which got deleted in modified file. This algorithm also finds the textual difference between two files and highlights those lines which are changed in modified file and which got deleted from original file.</p><p> </p>