Wednesday, December 11, 2013 - 2:00pm

This work studies the management of compliance with sustainability requirements in a supply chain in a business environment in which monitoring and legal enforcement of penalties is challenging. This situation is relevant to firms operating in developing countries where it is hard to make suppliers pay penalties for poor quality. Because buyers are concerned with... Read More

Wednesday, December 4, 2013 - 4:00pm

We study the performance of linear and piecewise-linear decision rules for adaptive optimization problems based only on the geometry of uncertainty sets. In particular, we show that Minkowski Symmetry and Banach-Mazur distance play a signicant role in determining the power of linear and piecewise-linear decision rules in adaptive optimization problems. We discuss the... Read More

Tuesday, December 3, 2013 - 3:00pm

Sampling large operational datasets such as ISP usage measurements can be effective for reducing storage requirements and execution time, while prolonging the useful life of the data for baselining and retrospective analysis. Sampling needs to mediate between the characteristics data and accuracy need of queries. This talk is about a cost-based formulation to express... Read More

Friday, November 15, 2013 - 2:30pm

A convex feasibility problem is concerned with finding a point in the intersection of convex sets. These problems are fundamental in optimization because any convex optimization problem can be cast in this form. Elementary algorithms are iterative algorithms for solving convex feasibility problems that involve simple operations such as matrix-vector multiplications,... Read More

Wednesday, November 13, 2013 - 4:00pm

The pooling problem is a challenging non-convex optimization problem that is motivated by refinery processes in the petroleum industry, and also finds application in other areas, such as waste water treatment, emissions regulation and agricultural industry among others. In this talk, we will present an analysis of mixed integer linear programming (MILP) based... Read More

Wednesday, November 6, 2013 - 4:00pm

Over the past 10 years there has been a growing body of work at the intersection of mathematical programming as commonly studied in Operations Research and constraint programming (CP) with its origins in Artificial Intelligence (AI) and programming languages. Much of CP's success in solving challenging combinatorial problems comes from the exploitation of inference as... Read More

Wednesday, October 30, 2013 - 4:00pm

A major challenge for machine learning in the next decade is the development of methods that continuously predict and learn from a stream of data. The scale of data and the non-stationary nature of the environment make the prevalent "learn from a batch of i.i.d. data" paradigm inadequate. In this talk, we will formally define the problem of sequential prediction as a (... Read More

Friday, October 18, 2013 - 2:30pm

Bayesian statistical models can be used to represent the beliefs of a decision-maker about an uncertain environment. For example, in revenue management, a seller formulates beliefs about customers' willingness to pay; in energy, we may have a belief about the suitability of a candidate location for a new wind farm. Conjugate priors model the evolution of these beliefs... Read More

Wednesday, October 16, 2013 - 4:00pm

For efficient and physically reliable numerical computations for time dependent differential equations it is very important to find positively invariant subsets of the state space and determine time step sizes of the numerical method that guarantee this. In this talk, we present the state-of-the-art theory and novel approaches beyond it to find positively invariant... Read More