Graduate Students Seminar

Speaker 1: Eswar Kammara

Title

Column Partition based Distributed Algorithms for Lasso

Abstract

This talk discusses column partition based distributed schemes for LASSO problems.  We are particularly interested in the cases where the number of (scalar) decision variables is much larger than the number of (scalar) measurements, and each agent has limited memory or computing capacity such that it only knows a small number of columns of a measurement matrix. The problems in consideration are densely coupled and cannot be formulated as separable convex programs. To overcome this difficulty, we consider their dual problems which are separable or locally coupled. Once a dual solution is attained, it is shown that a primal solution can be found from the dual of corresponding regularized BP-like problems under suitable exact regularization conditions. A wide range of existing distributed schemes can be exploited to solve the obtained dual problems. This yields two-stage column partition based distributed schemes for LASSO problems.

Speaker 2: Neha Agarwala

Title

Assessment of Individual- and Community-level Risks for COVID-19 Mortality in the US and Implications for Vaccine Distribution

Abstract

Equitable and effective risk-based allocations of scarce preventive resources, including early available vaccines will be essential to reduce COVID-19 illness and mortality for populations in the future. In this paper, we develop a risk calculator for COVID-19 mortality based on various socio-demographic, pre-existing behavioral and health conditions for the US adult population. We combine information from the UK-based OpenSAFELY study, with the mortality rates by age and ethnicity available across US states. In addition, we calibrate the tool to produce absolute risks for individuals using community level pandemic dynamics. We use the risk model and publicly available data on prevalence and co-occurrences of the risk-factors to project risk for the general adult population across 477 US cities and for the 65 years and older Medicare population across 3,113 US counties, respectively. Projections show that the model can identify relatively small fractions of the population which will lead to a disproportionately large number of deaths and thus will be useful for effectively targeting individuals for early vaccinations.