No description available.

The global objectives of this research are to develop a liver-specific magnetic resonance (MR) image registration and segmentation algorithms and to find highly correlated MR imaging features that help automatically score the severity of chronic liver disease (CLD). For a concise analysis of liver disease, time sequences of 3-D MR images should be preprocessed through an image registration to compensate for the patient motion, respiration, or tissue motion. To register contrast-enhanced MR image volume sequences, we propose a novel version of the demons algorithm that is based on a bi-directional local correlation coefficient (Bi-LCC) scheme. This scheme improves the speed at which a convergent sequence approaches to the optimum state and achieves the higher accuracy. Furthermore, the simple and parallelizable hierarchy of the Bi-LCC demons can be implemented on a graphics processing unit (GPU) using OpenCL. To automate segmentation of the liver parenchyma regions, an edge function-scaled region-based active contour (ESRAC), which hybridizes gradient and regional statistical information, with approximate partitions of the liver was proposed. Next, a significant purpose in grading liver disease is to assess the level of remaining liver function and to estimate regional liver function. On motion-corrected and segmented liver parenchyma regions, for quantitative analysis of the hepatic extraction of liver-specific MRI contrast agent, liver signal intensity change is evaluated from hepatobiliary phases (3-20 minutes), and parenchymal texture features are deduced from the equilibrium (3 minutes) phase. To build a classifier using texture features, a set of training input and output values, which is estimated by experts as a score of malignancy, trains the supervised learning algorithm using a multivariate normal distribution model and a maximum a posterior (MAP) decision rule. We validate the classifier by assessing the prediction accuracy with a set of testing data.

Thanks to the advanced technologies and social networks that allow the data to be widely shared among the Internet, there is an explosion of pervasive multimedia data, generating high demands of multimedia services and applications in various areas for people to easily access and manage multimedia data. Towards such demands, multimedia big data analysis has become an emerging hot topic in both industry and academia, which ranges from basic infrastructure, management, search, and mining to security, privacy, and applications. Within the scope of this dissertation, a multimedia big data analysis framework is proposed for semantic information management and retrieval with a focus on rare event detection in videos. The proposed framework is able to explore hidden semantic feature groups in multimedia data and incorporate temporal semantics, especially for video event detection. First, a hierarchical semantic data representation is presented to alleviate the semantic gap issue, and the Hidden Coherent Feature Group (HCFG) analysis method is proposed to capture the correlation between features and separate the original feature set into semantic groups, seamlessly integrating multimedia data in multiple modalities. Next, an Importance Factor based Temporal Multiple Correspondence Analysis (i.e., IF-TMCA) approach is presented for effective event detection. Specifically, the HCFG algorithm is integrated with the Hierarchical Information Gain Analysis (HIGA) method to generate the Importance Factor (IF) for producing the initial detection results. Then, the TMCA algorithm is proposed to efficiently incorporate temporal semantics for re-ranking and improving the final performance. At last, a sampling-based ensemble learning mechanism is applied to further accommodate the imbalanced datasets. In addition to the multimedia semantic representation and class imbalance problems, lack of organization is another critical issue for multimedia big data analysis. In this framework, an affinity propagation-based summarization method is also proposed to transform the unorganized data into a better structure with clean and well-organized information. The whole framework has been thoroughly evaluated across multiple domains, such as soccer goal event detection and disaster information management.

No description available.

Nowadays, smartphones are considered an essential part of our everyday lives. However, as the use of them increases so does their presence in vehicle environments. Prior studies have found that interacting with mobile devices while driving can cause cognitive, manual, and visual distractions. These distractions have a significant effect on driving performance and can result in serious accidents. As a solution to this, some applications, such as Spotify, has taken a different approach by developing a simplified version of their graphica linterface that is specifically designed for driving.The aim of this study is to examine what impacts simplified graphical interfaces has on the user experience for car drivers, focusing on Spotify and their car mode. In order to examine this, an online survey and a feature analysis was conducted. The result of this study shows that Spotify's car mode is missing a lot of features that are important to the users, causing them not to use it. These results highlight the challenge of providing all features that are expected by the users while creating a simpler interface.

[pt] A área de pesquisa em Montagem de Genomas tem evoluído rapidamente, adaptando-se às novas tecnologias de sequenciamento e modernos ambientes computacionais. Existem diversos softwares montadores que usam múltiplas abordagens, porém persiste o questionamento sobre a qualidade da montagem ao final do processo. Assim que uma montagem é finalizada, muitas medidas de qualidade podem ser geradas, a fim de que a montagem seja qualificada. Todavia, essas medidas apenas fornecem aos biólogos valores quantitativos acerca da montagem. Nós propomos nesta pesquisa um framework de domínio para o processo de análise de medidas pós montagem de genomas. Nosso objetivo é de prover a interpretação dos dados e avaliação da qualidade das montagens a partir do Framework. O Genome Assembly Analysis Framework (GAAF) foi projetado para trabalhar com espécies, montadores e medidas distintas. Para validar nossa proposta, foram realizados testes com o GAAF que permitem entender como o mesmo pode ser utilizado e de que maneira ele pode ser instanciado e/ou estendido. / [en] The Genome Assembly research area has quickly evolved, adapting to new sequencing technologies and modern computational environments. There exist many assembler software that consider multiple approaches. However, at the end of the process, one can always question the quality of assemblies. When an assembly is accomplished, some quality features may be generated, in order to qualify it. Nonetheless, the features do not directly tell one about assembly quality, but only bring to the biologists quantitative assembly descriptions. We propose a Domain Framework for the feature analysis process post-genome Assembly. Our goal is to enable data interpretation and assembly quality evaluation. The Genome Assembly Analysis Framework (GAAF) was designed to work with distinct species, assemblers and features. In order to validate our proposal, we have run a few practical experiments with GAAF, which make us understand the way it can be used, instantiated and extended.

In dieser Doktorarbeit wird Langsamkeit als unüberwachtes Lernprinzip in sensorischen Systemen untersucht. Dabei wird zwei Aspekten besondere Aufmerksamkeit gewidmet: der mathematischen Analyse von Slow Feature Analysis - einer Implementierung des Langsamkeitsprinzips - und der Frage, wie das Langsamkeitsprinzip biologisch umgesetzt werden kann. Im ersten Teil wird zunächst eine mathematische Theorie für Slow Feature Analysis entwickelt, die zeigt, dass die optimalen Funktionen für Slow Feature Analysis die Lösungen einer partiellen Differentialgleichung sind. Die Theorie erlaubt, das Verhalten komplizierter Anwendungen analytisch vorherzusagen und intuitiv zu verstehen. Als konkrete Anwendungen wird das Erlernen von Orts- und Kopfrichtungszellen, sowie von komplexen Zellen im primären visuellen Kortex vorgestellt. Im Rahmen einer technischen Anwendung werden die theoretischen Ergebnisse verwendet, um einen neuen Algorithmus für nichtlineare blinde Quellentrennung zu entwickeln und zu testen. Als Abschluss des ersten Teils wird die Beziehung zwischen dem Langsamkeitsprinzip und dem Lernprinzip der verhersagenden Kodierung mit Hilfe eines informationstheoretischen Ansatzes untersucht. Der zweite Teil der Arbeit befasst sich mit der Frage der biologischen Implementierung des Langsamkeitsprinzips. Dazu wird zunächst gezeigt, dass Spikezeit-abhängige Plastizität unter bestimmten Bedingungen als Implementierung des Langsamkeitsprinzips verstanden werden kann. Abschließend wird gezeigt, dass sich die Lerndynamik sowohl von gradientenbasiertem Langsamkeitslernen als auch von Spikezeit-abhängiger Plastizität mathematisch durch Reaktions-Diffusions-Gleichungen beschreiben lässt. / In this thesis, we investigate slowness as an unsupervised learning principle of sensory processing. Two aspects are given particular emphasis: (a) the mathematical analysis of Slow Feature Analysis (SFA) as one particular implementation of slowness learning and (b) the question, how slowness learning can be implemented in a biologically plausible fashion. In the first part of the thesis, we develop a mathematical framework for SFA and show that the optimal functions for SFA are the solutions of a partial differential eigenvalue problem. The theory allows (a) to make analytical predictions for the behavior of complicated applications and (b) an intuitive understanding of how the statistics of the input data are reflected in the optimal functions of SFA. The theory is applied to the learning of place and head-direction representations and to the learning of complex cell receptive fields as found in primary visual cortex. As a technical application, we use the theoretical results to develop and test a new algorithm for nonlinear blind source separation. The first part of the thesis is concluded by an information-theoretic analysis of the relation between slowness learning and predictive coding. In the second part of the thesis, we study the question, how slowness learning could be implemented in a biologically plausible manner. To this end, we first show that spike timing-dependent plasticity can under certain conditions be interpreted as an implementation of slowness learning. Finally, we show that both gradient-based slowness learning and spike timing-dependent plasticity lead to receptive field dynamics that can be described in terms of reaction-diffusion equations.

In dieser Doktorarbeit untersuchen wir zeitliche Langsamkeit als Prinzip für die Selbstorganisation des sensorischen Kortex sowie für computer-basierte Mustererkennung. Wir beginnen mit einer Einführung und Diskussion dieses Prinzips und stellen anschliessend den Slow Feature Analysis (SFA) Algorithmus vor, der das matemathisches Problem für diskrete Zeitreihen in einem endlich dimensionalen Funktionenraum löst. Im Hauptteil der Doktorarbeit untersuchen wir zeitliche Langsamkeit als Lernprinzip für rezeptive Felder im visuellen Kortex. Unter Verwendung von SFA werden Transformationsfunktionen gelernt, die, angewendet auf natürliche Bildsequenzen, möglichst langsam variierende Merkmale extrahieren. Die Funktionen können als nichtlineare raum-zeitliche rezeptive Felder interpretiert und mit Neuronen im primären visuellen Kortex (V1) verglichen werden. Wir zeigen, dass sie viele Eigenschaften von komplexen Zellen in V1 besitzen, nicht nur die primären, d.h. Gabor-ähnliche optimale Stimuli und Phaseninvarianz, sondern auch sekundäre, wie Richtungsselektivität, nicht-orthogonale Inhibition sowie End- und Seiteninhibition. Diese Resultate zeigen, dass ein einziges unüberwachtes Lernprinzip eine solche Mannigfaltigkeit an Eigenschaften begründen kann. Für die Analyse der mit SFA gelernten nichtlinearen Funktionen haben wir eine Reihe mathematischer und numerischer Werkzeuge entwickelt, mit denen man die quadratischen Formen als rezeptive Felder charakterisieren kann. Wir erweitern sie im weiteren Verlauf, um sie von allgemeinerem Interesse für theoretische und physiologische Modelle zu machen. Den Abschluss dieser Arbeit bildet die Anwendung des Prinzips der zeitlichen Langsamkeit auf Mustererkennungsprobleme. Die fehlende zeitliche Struktur in dieser Problemklasse erfordert eine Modifikation des SFA-Algorithmus. Wir stellen eine alternative Formulierung vor und wenden diese auf eine Standard-Datenbank von handgeschriebenen Ziffern an. / In this thesis we investigate the relevance of temporal slowness as a principle for the self-organization of the visual cortex and for technical applications. We first introduce and discuss this principle and put it into mathematical terms. We then define the slow feature analysis (SFA) algorithm, which solves the mathematical problem for multidimensional, discrete time series in a finite dimensional function space. In the main part of the thesis we apply temporal slowness as a learning principle of receptive fields in the visual cortex. Using SFA we learn the input-output functions that, when applied to natural image sequences, vary as slowly as possible in time and thus optimize the slowness objective. The resulting functions can be interpreted as nonlinear spatio-temporal receptive fields and compared to neurons in the primary visual cortex (V1). We find that they reproduce (qualitatively and quantitatively) many of the properties of complex cells in V1, not only the two basic ones, namely a Gabor-like optimal stimulus and phase-shift invariance, but also secondary ones like direction selectivity, non-orthogonal inhibition, end-inhibition and side-inhibition. These results show that a single unsupervised learning principle can account for a rich repertoire of receptive field properties. In order to analyze the nonlinear functions learned by SFA in our model, we developed a set of mathematical and numerical tools to characterize quadratic forms as receptive fields. We expand them in a successive chapter to be of more general interest for theoretical and physiological models. We conclude this thesis by showing the application of the temporal slowness principle to pattern recognition. We reformulate the SFA algorithm such that it can be applied to pattern recognition problems that lack of a temporal structure and present the optimal solutions in this case. We then apply the system to a standard handwritten digits database with good performance.

Feature extraction is an extremely important pre-processing step to pattern recognition, and machine learning problems. This thesis highlights how one can best extract features from the data in an exhaustively online and purely adaptive manner. The solution to this problem is given for both labeled and unlabeled datasets, by presenting a number of novel on-line learning approaches. Specifically, the differential equation method for solving the generalized eigenvalue problem is used to derive a number of novel machine learning and feature extraction algorithms. The incremental eigen-solution method is used to derive a novel incremental extension of linear discriminant analysis (LDA). Further the proposed incremental version is combined with extreme learning machine (ELM) in which the ELM is used as a preprocessor before learning. In this first key contribution, the dynamic random expansion characteristic of ELM is combined with the proposed incremental LDA technique, and shown to offer a significant improvement in maximizing the discrimination between points in two different classes, while minimizing the distance within each class, in comparison with other standard state-of-the-art incremental and batch techniques. In the second contribution, the differential equation method for solving the generalized eigenvalue problem is used to derive a novel state-of-the-art purely incremental version of slow feature analysis (SLA) algorithm, termed the generalized eigenvalue based slow feature analysis (GENEIGSFA) technique. Further the time series expansion of echo state network (ESN) and radial basis functions (EBF) are used as a pre-processor before learning. In addition, the higher order derivatives are used as a smoothing constraint in the output signal. Finally, an online extension of the generalized eigenvalue problem, derived from James Stone’s criterion, is tested, evaluated and compared with the standard batch version of the slow feature analysis technique, to demonstrate its comparative effectiveness. In the third contribution, light-weight extensions of the statistical technique known as canonical correlation analysis (CCA) for both twinned and multiple data streams, are derived by using the same existing method of solving the generalized eigenvalue problem. Further the proposed method is enhanced by maximizing the covariance between data streams while simultaneously maximizing the rate of change of variances within each data stream. A recurrent set of connections used by ESN are used as a pre-processor between the inputs and the canonical projections in order to capture shared temporal information in two or more data streams. A solution to the problem of identifying a low dimensional manifold on a high dimensional dataspace is then presented in an incremental and adaptive manner. Finally, an online locally optimized extension of Laplacian Eigenmaps is derived termed the generalized incremental laplacian eigenmaps technique (GENILE). Apart from exploiting the benefit of the incremental nature of the proposed manifold based dimensionality reduction technique, most of the time the projections produced by this method are shown to produce a better classification accuracy in comparison with standard batch versions of these techniques - on both artificial and real datasets.

In dieser Dissertation wird ein Modell des visuellen Systems untersucht, basierend auf dem Prinzip des unüberwachten Langsamkeitslernens und des SFA-Algorithmus (Slow Feature Analysis). Dieses Modell wird hier für die invariante Objekterkennung und verwandte Probleme eingesetzt. Das Modell kann dabei sowohl die zu Grunde liegenden diskreten Variablen der Stimuli extrahieren (z.B. die Identität des gezeigten Objektes) als auch kontinuierliche Variablen (z.B. Position und Rotationswinkel). Dabei ist es in der Lage, mit komplizierten Transformationen umzugehen, wie beispielsweise Tiefenrotation. Die Leistungsfähigkeit des Modells wird zunächst mit Hilfe von überwachten Methoden zur Datenanalyse untersucht. Anschließend wird gezeigt, dass auch die biologisch fundierte Methode des Verstärkenden Lernens (reinforcement learning) die Ausgabedaten unseres Modells erfolgreich verwenden kann. Dies erlaubt die Anwendung des Verstärkenden Lernens auf hochdimensionale visuelle Stimuli. Im zweiten Teil der Arbeit wird versucht, das hierarchische Modell mit Top-down Prozessen zu erweitern, speziell für die Rekonstruktion von visuellen Stimuli. Dabei setzen wir die Methode der Vektorquantisierung ein und verbinden diese mit einem Verfahren zum Gradientenabstieg. Die wesentlichen Komponenten der für unsere Simulationen entwickelten Software wurden in eine quelloffene Programmbibliothek integriert, in das ``Modular toolkit for Data Processing'''' (MDP). Diese Programmkomponenten werden im letzten Teil der Dissertation vorgestellt. / This thesis examines a model of the visual system, which is based on the principle of unsupervised slowness learning and using Slow Feature Analysis (SFA). We apply this model to the task of invariant object recognition and several related problems. The model not only learns to extract the underlying discrete variables of the stimuli (e.g., identity of the shown object) but also to extract continuous variables (e.g., position and rotational angles). It is shown to be capable of dealing with complex transformations like in-depth rotation. The performance of the model is first measured with the help of supervised post-processing methods. We then show that biologically motivated methods like reinforcement learning are also capable of processing the high-level output from the model. This enables reinforcement learning to deal with high-dimensional visual stimuli. In the second part of this thesis we try to extend the model with top-down processes, centered around the task of reconstructing visual stimuli. We utilize the method of vector quantization and combine it with gradient descent. The key components of our simulation software have been integrated into an open-source software library, the Modular toolkit for Data Processing (MDP). These components are presented in the last part of the thesis.