48670 research outputs found
Sort by
Martial\u27s Materials: Materiality In The Literary Epigram
Get PDFOriginating from inscribed epigram, concerning itself with occasional and satirical matters, and being written during the Flavian period, a time marked by efforts to catalogue and reframe Roman thought and tradition, Martial’s Epigrams understandably so are obsessed with the material world. Material objects, animate or inanimate, are at the center of interest of Martial’s poetry so much, that this dissertation suggests materiality as a fruitful lens through which Martial’s oeuvre as a whole can be approached. To do so, this study is structured into three avenues of investigation: sense perception, the (imagined) transformations of objects that are evoked through word plays, and a play with the representation of books and poets in poetry. This study finds that Martial often calls the very concept of materiality into question. This can occur e.g., when the poet portrays things that are not material, such as a smell, as palpable within his poetry. Elsewhere, the poet implicitly suggests a transformation of the legs of an individual by juxtaposing them with similarly shaped objects. Finally, the poet imagines concepts such as the greatness of an author as a material presence that can take up an entire room. Likewise, Martial alludes to an ubiquitous, dematerialized presence when he claims that “all of Rome reads me” or “I am in everyone’s pocket,” imagining himself as one with his book. The three chapters of my dissertation in conjunction shed light on how Martial’s material worldmaking suggests a coexistence of physical and conceptual materials that can both be captured by literary epigram. Literary epigram, thus, is fruitful for a reflection on matters of materiality: originating from being inscribed in stone, turned into ephemeral entertainment-pieces which lack coherency with one another and can be fragmented by the reader at will, literary epigram comes across as an anti-genre in which the material and the abstract lie close together
The Genetic Basis Of Natural Variation In The Timing Of Vegetative Phase Change In Arabidopsis Thaliana
No full textPlants undergo developmental changes that impact their form and function as they grow. Post-embryonic vegetative growth is characterized by two distinct developmental transitions: vegetative and reproductive phase changes. While natural variation in the reproductive transition is well studied, the transition between juvenile and adult vegetative growth — vegetative phase change — is often overlooked. Vegetative phase change is regulated by the microRNA, miR156, and its targets, SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPL) genes, and this pathway is conserved across land plants. However, the amount of natural variation in this transition, and the mechanisms regulating vegetative phase change in natural populations of Arabidopsis thaliana was unknown when I began my thesis. To better understand this problem, I assessed the phenotypic and genetic diversity in 100 natural accessions of A. thaliana. I found a wide range in the timing of vegetative phase change, that this transition is not correlated with reproductive phase change, and that miR156 and SPL gene expression were not associated with the observed phenotypic variation. These results imply that vegetative phase change is independent of other developmental transitions and that unknown genetic loci are involved in its regulation. I used two approaches to determine the genetic loci regulating natural variation in vegetative phase change: quantitative trait loci (QTL) mapping in the Shahdara accession and a genome-wide association study (GWAS) in a group of diverse accessions. QTL mapping in Shahdara revealed at least eight loci involved in the regulation of vegetative phase change, indicating regulation of vegetative phase change is complex and quantitative. GWAS for the timing of vegetative phase change uncovered several QTLs regulating vegetative phase change. The most significant polymorphisms were within a region that encodes a 24-nucelotide small RNA locus (smRNA), which delays abaxial trichome production. Together, this thesis provides the groundwork for future studies of the complex basis of natural variation in the timing of vegetative phase change in A. thaliana and has revealed novel genetic loci involved in the regulation of this transition. Additionally, my work has highlighted the importance of using natural variation to provide insights into the regulation of quantitative developmental pathways
Machine Learning On Large-Scale Graphs
Get PDFGraph neural networks (GNNs) are successful at learning representations from most types of network data but suffer from limitations in the case of large graphs. Challenges arise in the very design of the learning architecture, as most GNNs are parametrized by some matrix representation of the graph (e.g., the adjacency matrix) which can be hard to acquire when the network is large. Moreover, in many GNN architectures graph operations are defined through convolutional operations in the spectral domain. In this case, another obstacle is the obtention of the graph spectrum, which requires a costly matrix eigendecomposition.
Yet, large graphs can often be identified as being similar to each other in the sense that they share structural properties. We can thus expect that processing data supported on such graphs should yield similar results, which would mitigate the challenge of large size since we could then design GNNs for small graphs and transfer them to larger ones. In this thesis, I formalize this intuition and show that this graph transferability is possible when the graphs belong to the same family , where each family is identified by a different graphon.
A graphon is a function W(x,y) that describes a class of stochastic graphs with similar shape. One can think of the arguments (x,y) as the labels of a pair of nodes and of the graphon value W(x,y) as the probability of an edge between x and y. This yields a notion of a graph sampled from a graphon or, equivalently, a notion of a limit as the number of nodes in the sampled graph grows. Graphs sampled from a graphon almost surely share properties in the limit such as homomorphism densities which, in practice, implies that graphons identify families of networks that are similar in the sense that the density of certain motifs is preserved. This motivates the study of information processing on graphons as a way to enable information processing on large graphs.
The central component of a signal processing theory is a notion of shift that induces a class of linear filters with a spectral representation characterized by a Fourier transform (FT). In this thesis, we show that graphons induce a linear operator which can be used to define a shift and therefore graphon filters and the graphon FT. Building on the convergence properties of sequences of graphs and associated graph signals, it is then possible to show that for these sequences the graph FT converges to the graphon FT and that graph filter outputs converge to the outputs of the graphon filter with same coefficients. These theorems imply that for graphs that belong to certain families, graph Fourier analysis and graph filter design have well defined limits. In turn, these facts enable graph information processing on graphs with large number of nodes, since information processing pipelines designed for limit graphons can be applied to finite graphs.
We further define graphon neural networks (WNNs) by composing graphon filters banks with pointwise nonlinearities. WNNs are idealized limits which do not exist in practice, but they are a useful tool to understand the fundamental properties of GNNs. In particular, the sampling and convergence results derived for graphon filters canbe readily extended to WNNs, allowing to show that GNNs converge to WNNs as graphs converge to graphons. If two GNNs can be made arbitrarily close to the same WNN, then by a simple triangle inequality argument they can also be made arbitrarily close to one other. This result formalizes our intuition that GNNs are transferable between similar graphs.
A GNN can be trained on a moderate-scale graph and executed on a large-scale graph with a transferability error dominated by the inverse of the size of the smallest graph. Interestingly, this error increases with the variability of the spectral response of the convolutional filters, revealing a trade-off between transferability and spectral discriminability that is inherited from graph filters. In practice, this trade-off is less present in GNNs due to nonlinearities, which are able to scatter spectral components of the data to different parts of the eigenvalue spectrum where they can be discriminated. This explains why GNNs are more transferable than graph filters
Neural Synchrony In Successful Communication
Get PDFCommunicating our experiences to others relies on complex shared social, cultural, and psychological mechanisms. Research increasingly shows that shared neural mechanisms also play a role in the success of interpersonal communication. Synchronous activity in shared or complementary regions of the brain promotes emotional connections, cooperation, and memory between communicators. Regions of the brain involved in social and self-relevant information processes – (1) mentalizing, or thinking about the thoughts of others, and (2) self-relevance, or prospecting about the importance of information to the self – show synchrony in ways that correlate with communication outcomes. Synchrony can occur between two individuals, like speakers and their listeners, but it can also occur among a group of listeners, the audience. We use a form of neuroimaging called functional near-infrared spectroscopy to study neural activity as people tell and hear stories. First, we measure synchrony between storytellers and listeners. Chapter 2 shows that synchrony in mentalizing brain regions between a storyteller and her listeners predicts effective communication of emotional states. Next, we consider how synchrony across larger groups of audience members relates to successful communication. Chapter 3 demonstrates that an individual listener\u27s similarity to the average brain response in other audience members, in self-relevance processing regions, predicts the listener\u27s ability to authentically re-tell a story. Finally, extending this work, we also examine whether shared preferences predict neural synchrony in audience members. Chapter 4 integrates information about audience members’ individual preferences for content with audience-level neural synchrony. Within audiences of sports fans and theater lovers, self-reported content preferences predict behavioral liking for entertainment, but neural synchrony does not predict similar preferences in this case. Together these studies explore how synchrony between individuals predicts understanding and ability to transmit stories
Ethical Machine Learning: Fairness, Privacy, And The Right To Be Forgotten
Get PDFLarge-scale algorithmic decision making has increasingly run afoul of various social norms, laws, and regulations. A prominent concern is when a learned model exhibits discrimination against some demographic group, perhaps based on race or gender. Concerns over such algorithmic discrimination have led to a recent flurry of research on fairness in machine learning, which includes new tools for designing fair models, and studies the tradeoffs between predictive accuracy and fairness. We address algorithmic challenges in this domain.
Preserving privacy of data when performing analysis on it is not only a basic right for users, but it is also required by laws and regulations. How should one preserve privacy? After about two decades of fruitful research in this domain, differential privacy (DP) is considered by many the gold standard notion of data privacy. We focus on how differential privacy can be useful beyond preserving data privacy. In particular, we study the connection between differential privacy and adaptive data analysis.
Users voluntarily provide huge amounts of personal data to businesses such as Facebook, Google, and Amazon, in exchange for useful services. But a basic principle of data autonomy asserts that users should be able to revoke access to their data if they no longer find the exchange of data for services worthwhile. The right for users to request the erasure of personal data appears in regulations such as the Right to be Forgotten of General Data Protection Regulation (GDPR), and the California Consumer Privacy Act (CCPA). We provide algorithmic solutions to the the problem of removing the influence of data points from machine learning models
Unions And Social Capital In The United States
Get PDFSocial capital, as measured by social trust, pro-social values, or proxies capturing civic engagement, is a strong determinant of an area’s income, economic growth, and political institutions (Guiso, Sapienza and Zingales 2011). But in the US, it has declined significantly over the latter half of the twentieth century, with researchers and policymakers puzzling over why, and how it might be restored. In this dissertation, I propose and test a novel explanation for the decline in social capital—namely, the decline in unionization. Over much of the same period as the social capital decrease, union density decreased by over half, and workers lost “an important locus of social solidarity, a mechanism for mutual assistance and shared expertise” (Putnam 2000). To test this hypothesis, I estimate the effect of unionization on social capital, in two ways. The first is to combine data on NLRB unionization elections with data on commonly used proxies for social capital, and, in a modified RD-DD design, compare the change in social capital in areas that saw a close unionization victory to the change in areas that saw a close unionization loss. The second method is to use the PSID to estimate the effect of becoming unionized on an individual worker’s propensity for charitable donation. The results of the county-level analysis suggest that in the long term, unionization significantly increases voter turnout. However, the estimated effects on membership organizations are statistically insignificant. In an analysis of heterogeneous effects, I further find that the effect of unionization on total organizations is decreasing in a county’s per-capita income. Finally, the results of the individual-level analysis imply strong positive effects of becoming a union member on one’s family’s probability of donating to charity
Trans / Nation: Gender And Democracy In An Age Of Transition
No full textTrans / Nation analyzes two seemingly disparate uses of “transition”: first, to describe a person’s shift from one gender to another, and second, to narrate a nation’s political change through key terms like “democratization” and “development.” Rarely considered together, these invocations of transition form a unified history of state management from the 1970s to the present, masking neoliberal violence and promoting one “proper” path to prosperity for both individuals and nations.
Through case studies on Argentina, Chile, the Philippines, and Vietnam, Trans / Nation deconstructs slogans like “trans rights are human rights” by demonstrating that the recent recognition of trans people conceptualizes gender transition through the lens of national progress, whether from dictatorship to democracy in Chile or civil war to freedom in Vietnam. Using close readings of novels, film, drama, and archival materials, this project illustrates how nations exploit an ideology of transition to regulate internal populations, access new markets, and consolidate wealth for an elite class. Ultimately, Trans / Nation argues that gender transition and national transition are co-constitutive. Narratives of gender transition build on concepts of national autonomy, dictating what kinds of bodily changes are considered acceptable, while national transition narratives rely on the flexibility of gender to communicate a shift in state politics, whether or not significant change has occurred.
Trans / Nation assembles a wide range of narratives on trans and gender nonconforming people to build a history of refusal against ideologies of transition and connect strategies across nations and genres. Inspired by Édouard Glissant’s notion of opacity, Trans / Nation describes these tactics as a “trans opacity,” which aims to halt seemingly inevitable timelines of transition and expose forms of racist state discipline. In connecting Southeast Asia and the Southern Cone, trans opacity detaches transition from the nation, insisting instead on new forms of relationality and unexpected solidarities
Intentional Research Portfolio Growth in U.S. Research Universities During 2010–2018
No full textThe integrated mission of knowledge discovery, curation, application, and dissemination across nearly every field of inquiry is a uniquely American innovation. However, only a fraction of higher education institutions in the United States conduct the vast majority of academic research in this country. In fact, research expenditures in today’s academy follow a power law distribution and reflect a prestige hierarchy that has been in place for decades. This study investigated how leaders use intention in already research-intensive universities to martial necessary and sufficient resources and focus institutional research efforts to yield unrivaled advances in research expenditures and prestige. Specifically, this multisite case study employed qualitative research methods and a quantitative descriptive analysis to compare various strategies implemented by three high-performing public R1 universities that achieved outsized growth in their research portfolios during 2010–2018. Data were obtained through interviews with key university leaders, an analysis of publicly available institutional documents and websites, and examination of public datasets—principally, the NSF HERD survey. The universities in this study achieved robust sustained research portfolio growth by building directed momentum over time, accelerating a metaphorical flywheel of research growth with strategy ensembles that provided enhanced operational capacity which, in turn, altered and then bolstered institutional culture. New leadership was instrumental to success as they (1) leveraged fresh political capital and a new strategic plan to legitimate research excellence as an institutional priority, (2) installed a senior executive team to connect strategy to execution, and (3) grew the faculty ranks in strategic research domains. Using key research growth strategies, themselves commonplace across academe, these universities produced sustainable results at an institutional scale only once the intentionality behind these strategy ensembles led to the emergence of a culture of research excellence. Ultimately, three critical success factors for institutional advancement illustrate the difference between doing the right things and doing things right: leadership able to maintain persistent focus, the emergence of an enduring culture of research excellence, and the institution’s ability to execute on its determined strategy
Regulation of Anterior Gene Expression in the Caenorhabditis elegans Embryo
No full textPatterning of the anterior-posterior axis is fundamental to animal development. The Wnt pathway plays a major role in this process by activating the expression of posterior genes in animals from worms to humans. This observation raises the question of whether the Wnt pathway or other regulators control the expression of the many anterior-expressed genes. Using time-lapse laser confocal imaging of wild type and RNAi-treated C. elegans embryos, we found that the expression of five anterior-specific genes depends on the Wnt pathway effectors pop-1/TCF and sys-1/β-catenin. We focused further on one of these anterior genes, ref-2/ZIC, a conserved transcription factor expressed in multiple anterior lineages. Live imaging of ref-2 mutant embryos identified defects in cell division timing and position in anterior lineages. Cis-regulatory dissection identified three ref-2 transcriptional enhancers, one of which is necessary and sufficient for anterior-specific expression. This enhancer is activated by the T-box transcription factors TBX-37 and TBX-38, and surprisingly, concatemerized TBX-37/38 binding sites are sufficient to drive anterior-biased expression in a pop-1/TCF-dependent manner. Taken together, our results demonstrate that TCF can regulate an early-expressed anterior-biased gene in the C. elegans embryo through the binding site of another transcription factor
Reconstructing 3D Humans from Images
No full textThe past decade we have seen remarkable progress in Computer Vision, fueled by the recent advances in Deep Learning. Unsurprisingly, human perception has been the center of attention. We now have access to systems that can work remarkably well for traditional 2D tasks like segmentation or pose estimation. However, scaling this to 3D remains particularly challenging because of the inherent ambiguities and the scarcity of annotations. The goal of this dissertation is to describe our contributions towards automating 3D human reconstruction from images. First, we will explore the use of different representations for human mesh recovery, discuss their advantages and show how they can be useful for learning deformations beyond standard parametric body models. Next, motivated by the limited availability of annotated data, we will present a method that leverages a collaboration between regression and optimization methods to successfully address this. Subsequently, we will describe our work on modeling the ambiguities in 3D human reconstruction and demonstrate its usefulness for solving a variety of downstream tasks, such as human body model fitting. Last, we will move beyond single-person 3D pose estimation and show how we can scale our methods to work on challenging real-world scenes with multiple humans