Past Events

Join WNCG student Chao Jia for a PhD defense. Video data has increased dramatically in recent years due to the prevalence of handheld cameras. Such videos, however, are usually shakier compared to videos shot by tripod-mounted cameras or cameras with mechanical stabilizers. In addition, most handheld cameras use CMOS sensors. In a CMOS sensor camera, different rows in a frame are read/reset sequentially from top to bottom. When there is fast relative motion between the scene and the video camera, a frame can be distorted because each row was captured under a different 3D-to-2D projection.

CMOS technology scaling has fueled tremendous progress in electronics and has brought about system-on-chip (SoC) products with a broad impact on our society and economy. Technology scaling is very beneficial to increase the performance and density for digital signal processing, computation and memory. Analog and radio-frequency (RF) circuits remain the critical interfaces to connect the digital cores of SoCs to the physical world and need to satisfy increasing performance demands.

We consider generic optimization problems that can be formulated as minimizing the cost of a feasible solution w.x over a combinatorial feasible set F ⊂ {0, 1}^n. For these problems we describe a framework of risk-averse stochastic problems where the cost vector W has independent random components, unknown at the time of solution. A natural and important objective that incorporates risk in this stochastic setting is to look for a feasible solution whose stochastic cost has a small tail or a small convex combination of mean and standard deviation.

Community detection in networks is a key exploratory tool with applications ranging from identifying protein complexes in protein-protein interaction networks to scaling network computation by divide and conquer-type algorithms. Many clustering algorithms make heuristic choices which work well in practice, but their theoretical underpinnings are not well understood. For example, it is commonly believed that normalizing the network adjacency matrix improves the performance of Spectral Clustering.

In this discussion we consider load balancing on a large graph. Each edge has a unit of load that it wishes to distribute between its nodes in the most balanced way. For infinite graphs the corresponding load balancing problem exhibits nonuniqueness, related with role of boundary conditions in statistical mechanical models.  

Join WNCG Faculty Todd Humphreys, White Rose of Drachs' Captain Andrew Schofield and a panel of experts as they share the firsthand account of controlling a yacht with a custom-made GPS device. Must have SXSW Interactive Badge for entry.   For more information, visit the SXSW 2014 schedule. 

no results

Join Professor Christopher Jermaine from Rice University for a talk that describes the SimSQL System, a platform for writing and executing machine-learning codes. Since SimSQL is at its heart a relational database system, it is designed to support data independence. That is, the same declarative statistical inference codes can be used regardless of data set size, computer hardware, and physical data storage and distribution across machines.One concern is that a platform supporting data independence in this way will not perform well.

Modern sociotechnical systems are exemplified by individuals, teams and organizations working in concert with advanced and autonomous technologies to accomplish complex tasks. The challenge is to design these systems to capitalize on both human and machine capabilities by letting humans do what they do best and letting machines do what they do best to increase effectiveness and efficiency.

Rapid increase in the use of wireless services over the last two decades has lead the problem of the radio-frequency (RF) spectrum exhaustion. More specifically, due to this RF spectrum scarcity, additional RF bandwidth allocation, as utilized in the recent past, is not anymore a viable solution to fulfill the demand for more wireless applications and higher data rates.