Browsing by Subject "Methodology"

Differential privacy (DP) aims to design methods and algorithms that satisfy rigorous notions of privacy while simultaneously providing utility with valid statistical inference. More recently, an emphasis has been placed on combining notions of statistical utility with algorithmic approaches to address privacy risk in the presence of big data---with differential privacy emerging as a rigorous notion of risk. While DP provides strong guarantees for privacy, there are often tradeoffs regarding data utility and computational scalability. In this paper, we introduce a categorical data synthesizer that releases high-dimensional sparse histograms, illustrating its ability to overcome current limitations with data synthesizers in the current literature. Specifically, we combine a differential privacy algorithm---the stability based algorithm--- along with feature hashing, with allows for dimension reduction in terms of the histograms and Gibbs sampling. As a result, our proposed algorithm is differentially private, offers similar or better statistical utility and is scalable to large databases. In addition, we give an analytical result for the error caused by the stability based algorithm, which allows us to control the loss of utility. Finally, we study the behavior of our algorithm on both simulated and real data.

Quantification of metabolic fluxes has broad applications in studying metabolic physiology. Isotope tracing with heavy labeled carbon (13C labeled metabolic flux analysis, 13C-MFA) is a promising strategy due to its ability to compute metabolic fluxes from isotope tracing profiles without relying on assumptions such as metabolic objectives or enzyme kinetic parameters. However, most current 13C-MFA methods have limitations on model scope, algorithmic efficacy and user interaction, especially given the availability of modern mass spectrometry-based metabolomics techniques and computational resources. In this study, a new 13C-MFA framework is developed, with emphasis on flux resolution in larger-sized metabolic network. Novel simulation methods of isotope tracing data are also used to guide algorithm development. The new MFA methodology has been applied to isotope-labeled cultured cancer cell line and isotope-infused mice. In cultured cancer cell line, the new MFA framework enabled the discovery of long-term interaction from one-carbon metabolism to pentose phosphate pathway and TCA cycle. In isotope-infused mice, the new MFA framework directly measured systemic glucose and lactate contribution to TCA cycle under physiological condition, which confirms the conventional knowledge that glucose is the main direct energy source in body. Taken together the new MFA methodology will offer unprecedented opportunities for expanding research in metabolic physiology.

I argue that ethical convictions are crucial to the maintenance and transformation of social institutions. Moreover, since ethical convictions are sometimes corrigible and open to persuasive transformation, ethical persuasion can be a powerful source of social change. However, I observe that the dominant analytic techniques of the social sciences are ill equipped to understand the nature and import of ethical convictions, and even less well equipped to inform ethical persuasion. I argue this, in part, explains why social science research has often proved of little value in trying to address prominent social concerns.

This diagnosis raises a puzzle and a challenge. The puzzle is why some social scientists would wholly commit themselves to methods that cannot adequately deal with important dimensions of social structure. I show this is due to a misguided conception of science, one which seeks an "absolute perspective" that requires reducing or explaining away ethical convictions.

The challenge, once this vision of science is rejected in favor of a more pragmatic one, is 1) to understand the systematic limits of different methodological approaches and 2) to see how an account of ethics, rightly understood, can complement social scientific knowledge in service of better social outcomes.

I evaluate three dominant methodological approaches in the social sciences, namely, statistical modeling, formal modeling, and biological-behavioral research. Although all are useful within certain domains, I show that each has systematic limits relating to the dynamism of ethical convictions. I demonstrate how these methods can fail on their own terms and can blind researchers to important resources for social change, such as possibilities for persuasion.

Finally, I develop an account of the relationship between ethics, rationality, and persuasion drawing on the work of Hans-Georg Gadamer, Alasdair MacIntyre, and Charles Taylor. This account rejects prominent "scientific" attempts to explain ethical allegiances as biologically hardwired or structurally determined, and it further challenges accounts of ethical naturalism and pluralistic neutrality.

I conclude by illustrating the constructive role that ethical persuasion has played in a number of development projects, which help demonstrate my thesis that debates about visions of "the good" matter profoundly for human flourishing.