Georgia Institute of Technology College of Sciences

Statistical Process Control Charts are key tools in monitoring and controlling production processes to achieve conforming, high quality products. We will conduct a literature review on the Nonparametric Multivariate Statistical Process Control Charts to see what has been done in the area and how the methods have been applied.

In 1956 Mark Kac published his paper about the Foundation of Kinetic Theory in which he gave a mathematical, probabilistic description of a system of N particles colliding randomly. An interesting result that was found, though not causing any surprise, was the convergence to the stable equilibrium state. The question of the rate of the L2 convergence interested Kac and he conjectured that the spectral gap governing the convergence is uniformly bounded form below as N goes to infinity. While this was proved to be true, and even computed exactly, many situations show that the time scale of the convergence for very natural cases is proportional to N, while we would hope for an exponential decay. A different approach was considered, dealing with a more natural quantity, the entropy. In recent paper some advancement were made about evaluating the rate of change, and in 2003 Villani conjectured that the corresponding 'spectral gap', called the entropy production, is of order of 1/N. In our lecture we'll review the above topics and briefly discuss recently found results showing that the conjecture is essentially true.

We will discuss about the paper "An efficient algorithm for large-scale detection of protein families" by A Enright, S Van Dongen and C Ouzounis

We will start with a brief introduction to the broad area of machine learning, with the focus on empirical risk minimization methods and their connection to the theory of empirical processes. Using some results from our recent work with V. Koltchinskii, I will explain how sparsity affects the risk bounds.