Convex relaxation of mixture regression with efficient algorithms

We develop a convex relaxation of maximum a posteriori estimation of a mixture of regression models. Although our relaxation involves a semidefinite matrix variable, we reformulate the problem to eliminate the need for general semidefinite programming. In particular, we provide two reformulations that admit fast algorithms. The first is a max-min spectral reformulation exploiting quasi-Newton descent. The second is a min-min reformulation consisting of fast alternating steps of closed-form updates. We evaluate the methods against Expectation-Maximization in a real problem of motion segmentation from video data.

Substrate treatment and drying conditions effect on the properties of roll-to-roll gravure printed PEDOT:PSS thin films

The optical, structural and electrical properties of poly(3,4- ethylenedioxythiophene):poly(4-styrenesulfonic acid) (PEDOT:PSS) thin films printed by roll-to-roll gravure have been investigated. Corona treatment has been applied to enhance the adhesion of PEDOT:PSS on PolyEthylene Terephthalate (PET) web. It has been found that there was a stronger in-depth surface modification of PET with the increase of corona efficiency; however, the adhesion of PEDOT:PSS was not actually affected. Also, Spectroscopic Ellipsometry and Atomic Force Microscopy have been used to extract information on the mechanisms that define PEDOT:PSS properties. The increase of the drying temperature of the PEDOT:PSS films has been found to reduce the remaining water inside the films and lead to the decrease of the PEDOT:PSS particles size. © 2011 Elsevier B.V. All rights reserved.