Drexel University

About: Drexel University is a education organization based out in Philadelphia, Pennsylvania, United States. It is known for research contribution in the topics: Population & Poison control. The organization has 26770 authors who have published 51438 publications receiving 1949443 citations. The organization is also known as: Drexel & Drexel Institute.

Measuring Fairness in Ranked Outputs

Abstract: Ranking and scoring are ubiquitous We consider the setting in which an institution, called a ranker, evaluates a set of individuals based on demographic, behavioral or other characteristics The final output is a ranking that represents the relative quality of the individuals While automatic and therefore seemingly objective, rankers can, and often do, discriminate against individuals and systematically disadvantage members of protected groups This warrants a careful study of the fairness of a ranking scheme, to enable data science for social good applications, among othersIn this paper we propose fairness measures for ranked outputs We develop a data generation procedure that allows us to systematically control the degree of unfairness in the output, and study the behavior of our measures on these datasets We then apply our proposed measures to several real datasets, and detect cases of bias Finally, we show preliminary results of incorporating our ranked fairness measures into an optimization framework, and show potential for improving fairness of ranked outputs while maintaining accuracyThe code implementing all parts of this work is publicly available at https://githubcom/DataResponsibly/FairRank

Cellular Reservoirs of HIV-1 and their Role in Viral Persistence

Abstract: A major obstacle in human immunodeficiency virus type 1 (HIV-1) eradication is the ability of the virus to remain latent in a subpopulation of the cells it infects. Latently infected cells can escape the viral immune response and persist for long periods of time, despite the presence of successful highly active antiretroviral therapy (HAART). Given the appropriate stimulus, latently infected cells can reactivate and start producing infectious virions. The susceptibility of these cell populations to HIV-1, their life span, their proliferative capacity, and their ability to periodically produce infectious virus subsequent to alterations in cellular physiology and/or immunologic controls are critical issues which determine the contribution of these cells to viral persistence. Memory CD4+ T cells due to the long life span, which may be several years, and their ability to reactivate upon encounter with their cognate antigen or other stimulation, are considered a critical reservoir for maintenance of latent HIV-1 proviral DNA. Cells of the monocyte-macrophage lineage, which originate in the bone marrow (BM), are of particular importance in HIV-1 persistence due to their ability to cross the blood-brain barrier (BBB) and spread HIV-1 infection in the immunoprivileged central nervous system (CNS). Hematopoietic progenitor cells (HPCs) are also a potential HIV-1 reservoir, as several studies have shown that CD34+ HPCs carrying proviral DNA can be found in vivo in a subpopulation of HIV-1-infected patients. The ability of HPCs to proliferate and potentially generate clonal populations of infected cells of the monocyte-macrophage lineage may be crucial in HIV-1 dissemination. The contribution of these and other cell populations in HIV-1 persistence, as well as the possible strategies to eliminate latently infected cells are critically examined in this review.

Hiding Association Rules by Using Confidence and Support

Abstract: Large repositories of data contain sensitive information which must be protected against unauthorized access. Recent advances, in data mining and machine learning algorithms, have increased the disclosure risks one may encounter when releasing data to outside parties. A key problem, and still not sufficiently investigated, is the need to balance the confidentiality of the disclosed data with the legitimate needs of the data users. Every disclosure limitation method affects, in some way, and modifies true data values and relationships. In this paper, we investigate confidentiality issues of a broad category of rules, which are called association rules. If the disclosure risk of some of these rules are above a certain privacy threshold, those rules must be characterized as sensitive. Sometimes, sensitive rules should not be disclosed to the public since, among other things, they may be used for inferencing sensitive data, or they may provide business competitors with an advantage.

The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: Testing gravity with redshift-space distortions using the power spectrum multipoles

Abstract: We analyse the anisotropic clustering of the Baryon Oscillation Spectroscopic Survey (BOSS) CMASS Data Release 11 (DR11) sample, which consists of 690 827 galaxies in the redshift range 0.43 < z < 0.7 and has a sky coverage of 8498 deg^2. We perform our analysis in Fourier space using a power spectrum estimator suggested by Yamamoto et al. We measure the multipole power spectra in a self-consistent manner for the first time in the sense that we provide a proper way to treat the survey window function and the integral constraint, without the commonly used assumption of an isotropic power spectrum and without the need to split the survey into subregions. The main cosmological signals exploited in our analysis are the baryon acoustic oscillations and the signal of redshift space distortions, both of which are distorted by the Alcock–Paczynski effect. Together, these signals allow us to constrain the distance ratio D_V(z_eff)/r_s(z_d) = 13.89 ± 0.18, the Alcock–Paczynski parameter F_AP(z_eff) = 0.679 ± 0.031 and the growth rate of structure f (z_eff)σ_8(z_eff) = 0.419 ± 0.044 at the effective redshift z_eff = 0.57. We emphasize that our constraints are robust against possible systematic uncertainties. In order to ensure this, we perform a detailed systematics study against CMASS mock galaxy catalogues and N-body simulations. We find that such systematics will lead to 3.1 per cent uncertainty for fσ_8 if we limit our fitting range to k = 0.01–0.20 h Mpc^−1, where the statistical uncertainty is expected to be three times larger. We did not find significant systematic uncertainties for D_V/r_s or F_AP. Combining our data set with Planck to test General Relativity (GR) through the simple γ-parametrization, where the growth rate is given by $f(z) = \Omega ^{\gamma }_{\rm m}(z)$, reveals a ∼2σ tension between the data and the prediction by GR. The tension between our result and GR can be traced back to a tension in the clustering amplitude σ_8 between CMASS and Planck.