RICH AND EFFICIENT VISUAL DATA REPRESENTATION

Increasing the size of training data in many computer vision tasks has shown to be very effective. Using large scale image datasets (e.g. ImageNet) with simple learning techniques (e.g. linear classifiers) one can achieve state-of-the-art performance in object recognition compared to sophisticated learning techniques on smaller image sets. Semantic search on visual data has become very popular. There are billions of images on the internet and the number is increasing every day. Dealing with large scale image sets is intense per se. They take a significant amount of memory that makes it impossible to process the images with complex algorithms on single CPU machines. Finding an efficient image representation can be a key to attack this problem. A representation being efficient is not enough for image understanding. It should be comprehensive and rich in carrying semantic information. In this proposal we develop an approach to computing binary codes that provide a rich and efficient image representation. We demonstrate several tasks in which binary features can be very effective. We show how binary features can speed up large scale image classification. We present learning techniques to learn the binary features from supervised image set (With different types of semantic supervision; class labels, textual descriptions). We propose several problems that are very important in finding and using efficient image representation.

Sparse Signal Representation in Digital and Biological Systems

Theories of sparse signal representation, wherein a signal is decomposed as the sum of a small number of constituent elements, play increasing roles in both mathematical signal processing and neuroscience. This happens despite the differences between signal models in the two domains. After reviewing preliminary material on sparse signal models, I use work on compressed sensing for the electron tomography of biological structures as a target for exploring the efficacy of sparse signal reconstruction in a challenging application domain. My research in this area addresses a topic of keen interest to the biological microscopy community, and has resulted in the development of tomographic reconstruction software which is competitive with the state of the art in its field. Moving from the linear signal domain into the nonlinear dynamics of neural encoding, I explain the sparse coding hypothesis in neuroscience and its relationship with olfaction in locusts. I implement a numerical ODE model of the activity of neural populations responsible for sparse odor coding in locusts as part of a project involving offset spiking in the Kenyon cells. I also explain the validation procedures we have devised to help assess the model's similarity to the biology. The thesis concludes with the development of a new, simplified model of locust olfactory network activity, which seeks with some success to explain statistical properties of the sparse coding processes carried out in the network.

Interest Representation as a Clash of Unequal Allies

Organized interests do not have direct control over the fate of their policy agendas in Congress. They cannot introduce bills, vote on legislation, or serve on House committees. If organized interests want to achieve virtually any of their legislative goals they must rely on and work through members of Congress. As an interest group seeks to move its policy agenda forward in Congress, then, one of the most important challenges it faces is the recruitment of effective legislative allies. Legislative allies are members of Congress who “share the same policy objective as the group” and who use their limited time and resources to advocate for the group’s policy needs (Hall and Deardorff 2006, 76). For all the financial resources that a group can bring to bear as it competes with other interests to win policy outcomes, it will be ineffective without the help of members of Congress that are willing to expend their time and effort to advocate for its policy positions (Bauer, Pool, and Dexter 1965; Baumgartner and Leech 1998b; Hall and Wayman 1990; Hall and Deardorff 2006; Hojnacki and Kimball 1998, 1999). Given the importance of legislative allies to interest group success, are some organized interests better able to recruit legislative allies than others? This question has received little attention in the literature. This dissertation offers an original theoretical framework describing both when we should expect some types of interests to generate more legislative allies than others and how interests vary in their effectiveness at mobilizing these allies toward effective legislative advocacy. It then tests these theoretical expectations on variation in group representation during the stage in the legislative process that many scholars have argued is crucial to policy influence, interest representation on legislative committees. The dissertation uncovers pervasive evidence that interests with a presence across more congressional districts stand a better chance of having legislative allies on their key committees. It also reveals that interests with greater amounts of leverage over jobs and economic investment will be better positioned to win more allies on key committees. In addition, interests with a policy agenda that closely overlaps with the jurisdiction of just one committee in Congress are more likely to have legislative allies on their key committees than are interests that have a policy agenda divided across many committee jurisdictions. In short, how groups are distributed across districts, the leverage that interests have over local jobs and economic investment, and how committee jurisdictions align with their policy goals affects their influence in Congress.

The Drama of History: Representation and Revolutionaries in Haitian Theater, 1818-1907

Since the beginning of the Haitian theatrical tradition there has been an ineluctable dedication to the representation of Haitian history on stage. Given the rich theatrical archive about Haiti throughout the world, this study considers operas and plays written solely by Haitian playwrights. By delving into the works of Juste Chanlatte, Massillon Coicou, and Vendenesse Ducasse this study proposes a re-reading of Haitian theater that considers the stage as an innovative site for contesting negative and clichéd representations of the Haitian Revolution and its revolutionary leadership. A genre long mired in accusations of mimicking European literary forms, this study proposes a reevaluation of Haitian theater and its literary origins.