Event Category: Operations Research Seminar Series
-
Distributionally Robust Stochastic Optimization with Wasserstein Distance
12/1/16 | 4:15pm | E51-315 Reception to follow. Abstract: Consider an optimization problem under uncertainty. One may consider formulating it as a stochastic optimization problem. Often in such a setting, a “true” probability distribution may not be known. In fact, often the notion of a true probability distribution may not even be applicable. We consider an…
-
Modern Optimization in Observational Studies
11/10/16 | 4:15pm | E51-315 Reception to follow. Abstract: An observational study is an empirical investigation whose objective is to elucidate cause-and-effect relationships, yet wherein the assignment of individuals to the treatments being analyzed cannot be controlled by the researcher. Without the protections endowed by active assignment, observational studies are left susceptible to bias in the form…
-
Robust Allocation and Online Bundling in Online Advertising
11/3/16 | 4:15pm | E51-315 Reception to follow. Abstract: Online ads are delivered in a real-time fashion under uncertainty in an environment with strategic agents. Making such real-time (or online) decisions without knowing the future results in challenging stochastic optimization problems for ad selection and dynamic mechanism design problems for repeated auctions. In this talk, I will…
-
Allocation of Greenhouse Gas Emissions in Supply Chains
10/27/16 | 4:15pm | E51-315 Reception to follow. Abstract: In view of the challenges of meeting the goals set at the recent Climate Change Conference in Paris, it should be noted that the 2,500 largest global corporations account for more than 20% of global greenhouse gas (GHG) emissions, and that companies’ direct emissions average only 14% of…
-
High Dimensional Forecasting via Interpretable Vector Autoregression
10/20/16 | 4:15pm | E51-315 Reception to follow. Abstract: Vector autoregression (VAR) is a fundamental tool for modeling multivariate time series. However, as the number of component series is increased, the VAR model becomes overparameterized. Several authors have addressed this issue by incorporating regularized approaches, such as the lasso in VAR estimation. Traditional approaches address overparameterization by…
-
Approximations and Heuristics for a Class of Bilevel Programs
10/13/16 | 4:15pm | E51-315 Reception to follow. Abstract: Although bilevel programs are mostly intractable, some instances are amenable to efficient solution procedures, either exact or heuristic. In this presentation, I focus on the approximation of hard nonlinear instances by more tractable formulations in mixed integer linear format (MILP). The algorithmic framework is illustrated on a competitive…
-
A Semantic Approach for Estimating Consumer Content Preferences from Online Search Queries
10/6/16 | 4:15pm | E51-315 Reception to follow. Abstract: We develop an innovative topic model, Hierarchically Dual Latent Dirichlet Allocation (HDLDA), which not only identifies topics in search queries and webpages, but also how the topics in search queries relate to the topics in the corresponding top search results. Using the estimates from HDLDA, a consumer’s content…
-
On Wide Split Cuts for Mixed-Integer Programming
9/29/16 | 4:15pm | E51-315 Reception to follow. Abstract: In the classical theory for split cuts, the ‘width’ of a split set is always equal to one. We investigate cutting planes that arise when widening the associated disjunctions. This allows, e.g., to model non-contiguous domains of (integer) variables (or, stated differently, ‘holes’ in the domains). The validity…
-
Convex relaxations for the Optimal Flow Problem
9/22/16 | 4:15pm | E51-315 Reception to follow. Abstract: The AC optimal power flow (OPF) problem is a key optimization problem in the area of electrical power systems operations. We compare the strength of linear programing (LP), second order cone programming (SOCP) and semi-definite relaxations (SDP) of two formulations of the OPF formulation. Then we present a…
-
Optimal Trees
9/15/16 | 4:15pm | E51-315 Reception to follow. Abstract: Classification and Regression Trees (CART) were introduced by Breiman et al. in 1984 and is one of the most widely used methods in Machine Learning. As an indication of impact CART has attracted approximately 30,000 citations in Google Scholar. The method constructs a decision tree using a greedy…