We introduce the Stochastic Asynchronous Proximal Alternating Linearized Minimization (SAPALM) method, a block coordinate stochastic proximal-gradient method for solving nonconvex, nonsmooth optimization problems. SAPALM is the first asynchronous parallel optimization method that provably converges on a large class of nonconvex, nonsmooth problems. We prove that SAPALM matches the best known rates of convergence --- among synchronous or asynchronous methods --- on this problem class. We provide upper bounds on the number of workers for which we can expect to see a linear speedup, which match the best bounds known for less complex problems, and show that in practice SAPALM achieves this linear speedup. We demonstrate state-of-the-art performance on several matrix factorization problems.

We introduce PedSeg, a system for computing the boundaries of visually distinct geo-spatial objects. Knowing the precise boundaries or even bounding box approximations of geo-spatial objects is important for maintaining existing repositories such as gazetteers as well creating volunteered geographic information sources such as OpenStreetMap. PedSeg uses active contour image segmentation to determine an object's spatial extent from high-resolution overhead imagery. The novel aspect of this work is that the image segmentation is seeded with a GPS track acquired by simply walking around or otherwise traversing the approximate boundary of the target object (thus the prefix Ped). The technique is intended to be completely automated once the GPS track has been loaded into the system. This provides several advantages such as the user not needing to be skilled at image editing. In fact, the user does not even need to access, view, or interact with the imagery. The use of active contour image segmentation compensates for inaccuracies in the user acquired GPS tracks due to GPS error or the physical inaccessibility of the object boundary. We present examples of using PedSeg to compute the boundaries of a number of geo-spatial objects.