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.

Screening Diversity: Women & Work in Twenty-First-Century Popular Culture

Screening Diversity: Women and Work in Twenty-first Century Popular Culture explores contemporary representations of diverse professional women on screen. Audiences are offered successful women with limited concerns for feminism, anti-racism, or economic justice. I introduce the term viewsers to describe a group of movie and television viewers in the context of the online review platform Internet Movie Database (IMDb) and the social media platforms Twitter and Facebook. Screening Diversity follows their engagement in a representative sample of professional women on film and television produced between 2007 and 2015. The sample includes the television shows, Scandal, Homeland, VEEP, Parks and Recreation, and The Good Wife, as well as the movies, Zero Dark Thirty, The Proposal, The Heat, The Other Woman, I Don’t Know How She Does It, and Temptation. Viewsers appreciated female characters like Olivia (Scandal), and Maya (Zero Dark Thiry) who treated their work as a quasi-religious moral imperative. Producers and viewsers shared the belief that unlimited time commitment and personal identification were vital components of professionalism. However, powerful women, like The Proposal’s Margaret and VEEP’s Selina, were often called bitches. Some viewsers embraced bitch-positive politics in recognition of the struggles of women in power. Women’s disproportionate responsibility for reproductive labor, often compromises their ability to live up to moral standards of work. Unlike producers, viewsers celebrated and valued that labor. However, texts that included serious consideration of women as workers were frequently labelled chick flicks or soap operas. The label suggested that women’s labor issues were not important enough that they could be a topic of quality television or prestigious film, which bolstered the idea that workplace equality for women is not a problem in which the general public is implicated. Emerging discussions of racial injustice on television offered hope that these formations are beginning to shift.

‘DEVOTED TO THE INTERESTS OF HIS RACE���: BLACK OFFICEHOLDERS AND THE POLITICAL CULTURE OF FREEDOM IN WILMINGTON, NORTH CAROLINA, 1865-1877

This dissertation examines black officeholding in Wilmington, North Carolina, from emancipation in 1865 through 1876, when Democrats gained control of the state government and brought Reconstruction to an end. It considers the struggle for black office holding in the city, the black men who held office, the dynamic political culture of which they were a part, and their significance in the day-to-day lives of their constituents. Once they were enfranchised, black Wilmingtonians, who constituted a majority of the city’s population, used their voting leverage to negotiate the election of black men to public office. They did so by using Republican factionalism or what the dissertation argues was an alternative partisanship. Ultimately, it was not factional divisions, but voter suppression, gerrymandering, and constitutional revisions that made local government appointive rather than elective, Democrats at the state level chipped away at the political gains black Wilmingtonians had made.

Viral Bodies: Uncontrollable Blackness in Popular Culture and Everyday Life

Viral Bodies: Uncontrollable Blackness in Popular Culture and Everyday Life maps rapidly circulated performances of Blackness across visual media that collapse Black bodies into ubiquitous “things.” Throughout my dissertation, I use viral performance to describe the uncontrollable discursive circulation of bodies, their behaviors, and the ideas around them. In particular, viral performance is employed to describe the complicated ways that (mis)understandings of Black bodies spread and are often transformed into common-sense beliefs. As viral performances, Black bodies are often made more visible, while simultaneously becoming more opaque. This dissertation examines the recurrence of viral performances of Blackness in viral videos online, film, and photography/images. I argue that viral performances make products that reinscribe stereotypical notions of Blackness while also generating paths of alterity—which contradict the normalized clichés and provide desirable possibilities for Black performance. Viral Bodies forges a new dialogue between visual and aural technologies, performance, and larger historic discourses that script Black bodies as visually (and sonically) deviant subjects. I am interested in how technologies complicate the re-presentation of images, ideas, and ideologies—producing a necessity for new decipherings of performances of Blackness in popular culture and everyday life.

Multiraciality Enters the University: Mixed Race Identity and Knowledge Production in Higher Education

“Multiraciality Enters the University: Mixed Race Identity and Knowledge Production in Higher Education,” explores how the category of “mixed race��� has underpinned university politics in California, through student organizing, admissions debates, and the development of a new field of study. By treating the concept of privatization as central to both multiraciality and the neoliberal university, this project asks how and in what capacity has the discourses of multiracialism and the growing recognition of mixed race student populations shaped administrative, social, and academic debates at the state’s flagship universities—the University of California at Berkeley and Los Angeles. This project argues that the mixed race population symbolizing so-called “post-racial societies” is fundamentally attached to the concept of self-authorship, which can work to challenge the rights and resources for college students of color. Through a close reading of texts, including archival materials, policy and media debates, and interviews, I assert that the contemporary deployment of mixed race within the US academy represents a particularly post-civil rights development, undergirded by a genealogy of U.S. liberal individualism. This project ultimately reveals the pressing need to rethink ways to disrupt institutionalized racism in the new millennium.

Contested Play and Clean Water: McMillan Park, Race, and the Built Environment in Washington, D.C., 1900-1941

This study focuses on the intersection of the politics and culture of open public space with race relations in the United States from 1900 to 1941. The history of McMillan Park in Washington, D.C. serves as a lens to examine these themes. Ultimately, the park’s history, as documented in newspapers, interviews, reports, and photographs, reveals how white residents attempted to protect their dominance in a racial hierarchy through the control of both the physical and cultural elements of public recreation space. White use of discrimination through seemingly neutral desires to protect health, safety, and property values, establishes a congruence with their defense of residential property. Without similar access to legal methods, African Americans acted through direct action in gaps of governmental control. Their use of this space demonstrates how African-American residents of Washington and the United States contested their race, recreation, and spatial privileges in the pre-World War II era.

Generating Discriminative Objective Proposals via Submodular Ranking

Object recognition has long been a core problem in computer vision. To improve object spatial support and speed up object localization for object recognition, generating high-quality category-independent object proposals as the input for object recognition system has drawn attention recently. Given an image, we generate a limited number of high-quality and category-independent object proposals in advance and used as inputs for many computer vision tasks. We present an efficient dictionary-based model for image classification task. We further extend the work to a discriminative dictionary learning method for tensor sparse coding. In the first part, a multi-scale greedy-based object proposal generation approach is presented. Based on the multi-scale nature of objects in images, our approach is built on top of a hierarchical segmentation. We first identify the representative and diverse exemplar clusters within each scale. Object proposals are obtained by selecting a subset from the multi-scale segment pool via maximizing a submodular objective function, which consists of a weighted coverage term, a single-scale diversity term and a multi-scale reward term. The weighted coverage term forces the selected set of object proposals to be representative and compact; the single-scale diversity term encourages choosing segments from different exemplar clusters so that they will cover as many object patterns as possible; the multi-scale reward term encourages the selected proposals to be discriminative and selected from multiple layers generated by the hierarchical image segmentation. The experimental results on the Berkeley Segmentation Dataset and PASCAL VOC2012 segmentation dataset demonstrate the accuracy and efficiency of our object proposal model. Additionally, we validate our object proposals in simultaneous segmentation and detection and outperform the state-of-art performance. To classify the object in the image, we design a discriminative, structural low-rank framework for image classification. We use a supervised learning method to construct a discriminative and reconstructive dictionary. By introducing an ideal regularization term, we perform low-rank matrix recovery for contaminated training data from all categories simultaneously without losing structural information. A discriminative low-rank representation for images with respect to the constructed dictionary is obtained. With semantic structure information and strong identification capability, this representation is good for classification tasks even using a simple linear multi-classifier.