Learning from Geometry

Subspaces and manifolds are two powerful models for high dimensional signals. Subspaces model linear correlation and are a good fit to signals generated by physical systems, such as frontal images of human faces and multiple sources impinging at an antenna array. Manifolds model sources that are not linearly correlated, but where signals are determined by a small number of parameters. Examples are images of human faces under different poses or expressions, and handwritten digits with varying styles. However, there will always be some degree of model mismatch between the subspace or manifold model and the true statistics of the source. This dissertation exploits subspace and manifold models as prior information in various signal processing and machine learning tasks.

A near-low-rank Gaussian mixture model measures proximity to a union of linear or affine subspaces. This simple model can effectively capture the signal distribution when each class is near a subspace. This dissertation studies how the pairwise geometry between these subspaces affects classification performance. When model mismatch is vanishingly small, the probability of misclassification is determined by the product of the sines of the principal angles between subspaces. When the model mismatch is more significant, the probability of misclassification is determined by the sum of the squares of the sines of the principal angles. Reliability of classification is derived in terms of the distribution of signal energy across principal vectors. Larger principal angles lead to smaller classification error, motivating a linear transform that optimizes principal angles. This linear transformation, termed TRAIT, also preserves some specific features in each class, being complementary to a recently developed Low Rank Transform (LRT). Moreover, when the model mismatch is more significant, TRAIT shows superior performance compared to LRT.

The manifold model enforces a constraint on the freedom of data variation. Learning features that are robust to data variation is very important, especially when the size of the training set is small. A learning machine with large numbers of parameters, e.g., deep neural network, can well describe a very complicated data distribution. However, it is also more likely to be sensitive to small perturbations of the data, and to suffer from suffer from degraded performance when generalizing to unseen (test) data.

From the perspective of complexity of function classes, such a learning machine has a huge capacity (complexity), which tends to overfit. The manifold model provides us with a way of regularizing the learning machine, so as to reduce the generalization error, therefore mitigate overfiting. Two different overfiting-preventing approaches are proposed, one from the perspective of data variation, the other from capacity/complexity control. In the first approach, the learning machine is encouraged to make decisions that vary smoothly for data points in local neighborhoods on the manifold. In the second approach, a graph adjacency matrix is derived for the manifold, and the learned features are encouraged to be aligned with the principal components of this adjacency matrix. Experimental results on benchmark datasets are demonstrated, showing an obvious advantage of the proposed approaches when the training set is small.

Stochastic optimization makes it possible to track a slowly varying subspace underlying streaming data. By approximating local neighborhoods using affine subspaces, a slowly varying manifold can be efficiently tracked as well, even with corrupted and noisy data. The more the local neighborhoods, the better the approximation, but the higher the computational complexity. A multiscale approximation scheme is proposed, where the local approximating subspaces are organized in a tree structure. Splitting and merging of the tree nodes then allows efficient control of the number of neighbourhoods. Deviation (of each datum) from the learned model is estimated, yielding a series of statistics for anomaly detection. This framework extends the classical {\em changepoint detection} technique, which only works for one dimensional signals. Simulations and experiments highlight the robustness and efficacy of the proposed approach in detecting an abrupt change in an otherwise slowly varying low-dimensional manifold.

A look into diversity in the radio and music industry in Spain

UNESCO’s approval of the Convention on the Protection and Promotion of the Diversity of Cultural Expressions (UNESCO, 2005) has been an important element in catalyzing any attempt to measure the diversity of cultural industries (UIS, 2011). Within this framework, this article analyzes the relations between the music and radio industries in Spain from a critical perspective through the analysis of available data on recorded music offer and consumption (sales lists, radio-formula lists, the characteristics of the phonographic and radio markets) in different key moments due to the emergence of new formats and devices (CDS, Mp3, Internet).The main goal of this work is to study the evolution of the Spanish record market in terms of diversity from the end of the 1970s to the present, through the study of radio music hits lists and, the business structure of the phonographic and radio sectors, and phonograms top sales

Who is in charge? An L2 Discourse Intonation Study on Four Prosodic Parameters to Exert the Pragmatic Function of Dominance and Control in the Context of L2 Non-specialist Public Speaking

This paper reports the findings from a study of the learning of English intonation by Spanish speakers within the discourse mode of L2 oral presentation. The purpose of this experiment is, firstly, to compare four prosodic parameters before and after an L2 discourse intonation training programme and, secondly, to confirm whether subjects, after the aforementioned L2 discourse intonation training, are able to match the form of these four prosodic parameters to the discourse-pragmatic function of dominance and control. The study designed the instructions and tasks to create the oral and written corpora and Brazil’s Pronunciation for Advanced Learners of English was adapted for the pedagogical aims of the present study. The learners’ pre- and post-tasks were acoustically analysed and a pre / post- questionnaire design was applied to interpret the acoustic analysis. Results indicate most of the subjects acquired a wider choice of the four prosodic parameters partly due to the prosodically-annotated transcripts that were developed throughout the L2 discourse intonation course. Conversely, qualitative and quantitative data reveal most subjects failed to match the forms to their appropriate pragmatic functions to express dominance and control in an L2 oral presentation.

A parallel pattern for iterative stencil + reduce

We advocate the Loop-of-stencil-reduce pattern as a means of simplifying the implementation of data-parallel programs on heterogeneous multi-core platforms. Loop-of-stencil-reduce is general enough to subsume map, reduce, map-reduce, stencil, stencil-reduce, and, crucially, their usage in a loop in both data-parallel and streaming applications, or a combination of both. The pattern makes it possible to deploy a single stencil computation kernel on different GPUs. We discuss the implementation of Loop-of-stencil-reduce in FastFlow, a framework for the implementation of applications based on the parallel patterns. Experiments are presented to illustrate the use of Loop-of-stencil-reduce in developing data-parallel kernels running on heterogeneous systems.

A series of performances in 'Translating Reviews from 1990: Art Forum'

Participation in group exhibition themed around the 25th anniversary of the Elba Benitez Gallery in Madrid. My work comprised a series of performances in which I translated reviews from the magazine Art Forum from 1990. The performances took place in various locations in London, throughout the run of the exhibition, and were streamed live to an iPad in the gallery in Madrid. I made audio visual recordings of the performances via the streaming media, which located me as the performer alongside the viewers in a single split image. These recordings were then archived in a shared folder held between the gallery and me, and which visitors to the exhibition could access when a performance was not taking place. The work extends my concerns with translation and performance, and with a consideration of how the mechanism of the gallery and the exhibition might be used to generate innovative viewing engagements facilitated by technology. The work also attempts to develop thinking and practice around the relationship between art works and their documentation - in this case the documentation and even its potential for distribution is generated as the work comes into being. The exhibition included works by Ignasi Aballí, Armando Andrade Tudela,Lothar Baumgarten, Carlos Bunga, Cabello/Carceller, Juan Cruz, Gintaras Didžiapetris, Fernanda Fragateiro, Hreinn Fridfinnsson, Carlos Garaicoa,Mario García Torres, David Goldblatt, Cristina Iglesias,Ana Mendieta, Vik Muniz, Ernesto Neto, Francisco Ruiz de Infante,Alexander Sokurov, Francesc Torres and Valentín Vallhonrat.

Fragmentation versus convergence: University students in Brussels and the consumption of TV series on the Internet

info:eu-repo/semantics/published

A Multimedia Video Player Based of FFmpeg

With the rapid development of Internet technologies, video and audio processing are among the most important parts due to the constant requirements of high quality media contents. Along with the improvement of network environment and the hardware equipment, this demand is becoming more and more imperious, people prefer high quality videos and audios as well as the net streaming media resources. FFmpeg is a set of open source program about the A/V decoding. Many commercial players use FFmpeg as their displaying cores. This paper designed a simple and easy-to-use video player based on FFmpeg. The first part is about the basic theories and related knowledge of video displaying, including some concepts like data formats, streaming media data, video coding and decoding. In a word, the realization of the video player depend on the a set of video decoding process. The general idea about the process is to get the video packets from the Internet, to read the related protocols and de-encapsulate the protocols, to de-encapsulate the packaging data and to get encoded formats data, to decode them to pixel data that can be displayed directly through graphics cards. During the coding and decoding process, there could be different degrees of data losing, which is called lossy compression, but it usually does not influence the quality of user experiences. The second part is about the principle of the FFmpeg decoding process, that is one of the key point of the paper. In this project, FFmpeg is used for the main decoding task, by call some main functions and structures from FFmpeg class libraries, packaging video formats could be transfer to pixel data, after getting the pixel data, SDL is used for the displaying process. The third part is about the SDL displaying flow. Similarly, it would invoke some important displaying functions from SDL class libraries to realize the function, though SDL is able to do not only displaying task, but also many other game playing process. After that, a independent video displayer is completed, it is provided with all the key function of a player. The fourth part make a simple users interface for the player based on the MFC program, it enable the player could be used by most people. At last, in consideration of the mobile Internet’s blossom, people nowadays can hardly ever drop their mobile phones, there is a brief introduction about how to transplant the video player to Android platform which is one of the most used mobile systems.

New Library-curated Faculty Profile Platform Launches Through bepress

As faculty needs evolve and become increasingly digital, libraries are feeling the pressure to provide relevant new services. At the same time, faculty members are struggling to create and maintain their professional reputations online. We at bepress are happy to announce the new SelectedWorks, the fully hosted, library-curated faculty profile platform that positions the library to better support faculty as well as the institution at large. Beverly Lysobey, Digital Commons and Resource Management Librarian, at Sacred Heart University, says: “Both faculty and administration have been impressed with the services we provide through SelectedWorks; we’re able to show how much our faculty really publishes, and it’s great for professors to get that recognition. We’ve had several faculty members approach us for help making sure their record was complete when they were up for tenure, and we’ve even found articles that authors themselves no longer had access to.” With consistent, organized, institution-branded profiles, SelectedWorks increases campus-wide exposure and supports the research mission of the university. As the only profile platform integrated with the fully hosted Digital Commons suite of publishing and repository services, it also ensures that the institution retains management of its content. Powerful integration with the Digital Commons platform lets the home institution more fully capture the range of scholarship produced on campus, and hosted services facilitate resource consolidation and reduces strain on IT. The new SelectedWorks features a modern, streamlined design that provides compelling display options for the full range of faculty work. It beautifully showcases streaming media, images, data, teaching materials, books – any type of content that researchers now produce as part of their scholarship. Detailed analytics tools let authors and librarians measure global readership and track impact for a variety of campus stakeholders: authors can see the universities, agencies, and businesses that are reading their work, and can easily export reports to use in tenure and promotion dossiers. Janelle Wertzbeger, Assistant Dean and Director of Scholarly Communications at Gettysburg College’s Musselman Library, says, “The new author dashboard maps and enhanced readership are SO GOOD. Every professor up for promotion & tenure should use them!” And of course, SelectedWorks is fully backed by the continual efforts of the bepress development team to provide maximum discoverability to search engines, increasing impact for faculty and institutions alike: Reverend Edward R. Udovic, Vice President for Teaching and Learning Resources at DePaul University, says, “In the last several months downloads of my scholarship from my [SelectedWorks] site have far surpassed the total distribution of all my work in the previous twenty five years.”

Real-time analytics on large dynamic graphs

In today's fast-paced and interconnected digital world, the data generated by an increasing number of applications is being modeled as dynamic graphs. The graph structure encodes relationships among data items, while the structural changes to the graphs as well as the continuous stream of information produced by the entities in these graphs make them dynamic in nature. Examples include social networks where users post status updates, images, videos, etc.; phone call networks where nodes may send text messages or place phone calls; road traffic networks where the traffic behavior of the road segments changes constantly, and so on. There is a tremendous value in storing, managing, and analyzing such dynamic graphs and deriving meaningful insights in real-time. However, a majority of the work in graph analytics assumes a static setting, and there is a lack of systematic study of the various dynamic scenarios, the complexity they impose on the analysis tasks, and the challenges in building efficient systems that can support such tasks at a large scale. In this dissertation, I design a unified streaming graph data management framework, and develop prototype systems to support increasingly complex tasks on dynamic graphs. In the first part, I focus on the management and querying of distributed graph data. I develop a hybrid replication policy that monitors the read-write frequencies of the nodes to decide dynamically what data to replicate, and whether to do eager or lazy replication in order to minimize network communication and support low-latency querying. In the second part, I study parallel execution of continuous neighborhood-driven aggregates, where each node aggregates the information generated in its neighborhoods. I build my system around the notion of an aggregation overlay graph, a pre-compiled data structure that enables sharing of partial aggregates across different queries, and also allows partial pre-computation of the aggregates to minimize the query latencies and increase throughput. Finally, I extend the framework to support continuous detection and analysis of activity-based subgraphs, where subgraphs could be specified using both graph structure as well as activity conditions on the nodes. The query specification tasks in my system are expressed using a set of active structural primitives, which allows the query evaluator to use a set of novel optimization techniques, thereby achieving high throughput. Overall, in this dissertation, I define and investigate a set of novel tasks on dynamic graphs, design scalable optimization techniques, build prototype systems, and show the effectiveness of the proposed techniques through extensive evaluation using large-scale real and synthetic datasets.

Probabilistic Models for Scalable Knowledge Graph Construction

In the past decade, systems that extract information from millions of Internet documents have become commonplace. Knowledge graphs -- structured knowledge bases that describe entities, their attributes and the relationships between them -- are a powerful tool for understanding and organizing this vast amount of information. However, a significant obstacle to knowledge graph construction is the unreliability of the extracted information, due to noise and ambiguity in the underlying data or errors made by the extraction system and the complexity of reasoning about the dependencies between these noisy extractions. My dissertation addresses these challenges by exploiting the interdependencies between facts to improve the quality of the knowledge graph in a scalable framework. I introduce a new approach called knowledge graph identification (KGI), which resolves the entities, attributes and relationships in the knowledge graph by incorporating uncertain extractions from multiple sources, entity co-references, and ontological constraints. I define a probability distribution over possible knowledge graphs and infer the most probable knowledge graph using a combination of probabilistic and logical reasoning. Such probabilistic models are frequently dismissed due to scalability concerns, but my implementation of KGI maintains tractable performance on large problems through the use of hinge-loss Markov random fields, which have a convex inference objective. This allows the inference of large knowledge graphs using 4M facts and 20M ground constraints in 2 hours. To further scale the solution, I develop a distributed approach to the KGI problem which runs in parallel across multiple machines, reducing inference time by 90%. Finally, I extend my model to the streaming setting, where a knowledge graph is continuously updated by incorporating newly extracted facts. I devise a general approach for approximately updating inference in convex probabilistic models, and quantify the approximation error by defining and bounding inference regret for online models. Together, my work retains the attractive features of probabilistic models while providing the scalability necessary for large-scale knowledge graph construction. These models have been applied on a number of real-world knowledge graph projects, including the NELL project at Carnegie Mellon and the Google Knowledge Graph.

Internationalization venture analysis of NMusic to the Spanish market

Field lab: Entrepreneurial and innovative ventures

REGULATION OF SHAPE DYNAMICS AND ACTIN POLYMERIZATION DURING COLLECTIVE CELL MIGRATION

This thesis aims to understand how cells coordinate their motion during collective migration. As previously shown, the motion of individually migrating cells is governed by wave-like cell shape dynamics. The mechanisms that regulate these dynamic behaviors in response to extracellular environment remain largely unclear. I applied shape dynamics analysis to Dictyostelium cells migrating in pairs and in multicellular streams and found that wave-like membrane protrusions are highly coupled between touching cells. I further characterized cell motion by using principle component analysis (PCA) to decompose complex cell shape changes into a serial shape change modes, from which I found that streaming cells exhibit localized anterior protrusion, termed front narrowing, to facilitate cell-cell coupling. I next explored cytoskeleton-based mechanisms of cell-cell coupling by measuring the dynamics of actin polymerization. Actin polymerization waves observed in individual cells were significantly suppressed in multicellular streams. Streaming cells exclusively produced F-actin at cell-cell contact regions, especially at cell fronts. I demonstrated that such restricted actin polymerization is associated with cell-cell coupling, as reducing actin polymerization with Latrunculin A leads to the assembly of F-actin at the side of streams, the decrease of front narrowing, and the decoupling of protrusion waves. My studies also suggest that collective migration is guided by cell-surface interactions. I examined the aggregation of Dictyostelim cells under distinct conditions and found that both chemical compositions of surfaces and surface-adhesion defects in cells result in altered collective migration patterns. I also investigated the shape dynamics of cells suspended on PEG-coated surfaces, which showed that coupling of protrusion waves disappears on touching suspended cells. These observations indicate that collective migration requires a balance between cell-cell and cell-surface adhesions. I hypothesized such a balance is reached via the regulation of cytoskeleton. Indeed, I found cells actively regulate cytoskeleton to retain optimal cell-surface adhesions on varying surfaces, and cells lacking the link between actin and surfaces (talin A) could not retain the optimal adhesions. On the other hand, suspended cells exhibited enhanced actin filament assembly on the periphery of cell groups instead of in cell-cell contact regions, which facilitates their aggregation in a clumping fashion.

SC REVENUE RULING #16-5

This ruling proclaims that charges paid by a customer for streaming television programs, movies, music, and other similar content are charges for communication services and are therefore subject to South Carolina sales and use tax whether paid for as part of a subscription service, per item, or per event.

Arquitetura de rede com suporte à mobilidade de fluxo IP e ao gerenciamento de Mobilidade Distribuído (DMM)

Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Elétrica, 2016.

Dynamic Multi-Objective Optimization With jMetal and Spark: a Case Study

Technologies for Big Data and Data Science are receiving increasing research interest nowadays. This paper introduces the prototyping architecture of a tool aimed to solve Big Data Optimization problems. Our tool combines the jMetal framework for multi-objective optimization with Apache Spark, a technology that is gaining momentum. In particular, we make use of the streaming facilities of Spark to feed an optimization problem with data from different sources. We demonstrate the use of our tool by solving a dynamic bi-objective instance of the Traveling Salesman Problem (TSP) based on near real-time traffic data from New York City, which is updated several times per minute. Our experiment shows that both jMetal and Spark can be integrated providing a software platform to deal with dynamic multi-optimization problems.