It is expected that in future battlefield and disaster response situations, teams of unmanned vehicles, remote sensors, and people will work together, using an ad hoc wireless network to communicate. Unmanned aerial vehicles (UAVs) acting as mobile communication nodes have been proposed to supplement the ground-based network, which may be limited due to a noisy environment with obstructions and interference. In this talk I will describe graph theoretic and network flow problems of where to position the UAVs in order to facilitate communication within groups of nodes, and present several techniques for solving these problems.
(Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.)
How does the brain encode semantic information? For instance, how is your brain activated when you are thinking of the concept "dog" versus "carrot"? In this talk, I will present a generative computational model of fMRI brain activations associated with semantic information of concrete nouns. The model posits some intermediate semantic features underlying these fMRI brain activations. I will also present results testing the predictive ability of the model when used with several kinds of intermediate semantic features derived from large text corpus data.
Joint work with Tom Mitchell, Kai-min Chang, and Mark Palatucci.
(Presented in partial fulfillment of the CSD Speaking Skills Requirement.)
Given a terabyte click log, can we build an efficient and effective click model? It is commonly believed that web search click logs are a gold mine for search business, because they reflect users' preference over web documents presented by the search engine. Click models provide a principled approach to inferring user-perceived relevance of web documents, which can be leveraged in numerous applications in search businesses. Due to the huge volume of click data, scalability is a must. I will present the click chain model, which is based on a solid, Bayesian framework. It is both scalable and incremental, perfectly meeting the computational challenges imposed by the voluminous click logs that constantly grow.
Joint work with Chao Liu, Anitha Kannan, Tom Minka, Michael Taylor, Yi-Min Wang and Christos Faloutsos.
(Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.)
This presentation reports on a virtual experiment recently conducted to develop fresh theory and hypotheses on the organization behavioral aspects of post-merger integration. Experiments were conducted using the Construct model of social interaction to explore the effects of the pre-merger characteristics of the two organizations on the post-merger process. This study investigated the broad construct of organizational complexity as it applies to organization and work unit size and the effect on knowledge-based organization performance. Results suggest interesting hypotheses for further empirical investigation.
Policies for protecting sensitive information are often written in natural language and enforced using access control lists. These mechanisms are not only difficult for administrators but also error prone. Proof-carrying authorization (PCA) provides an alternate, logic-based, rigorous enforcement for policies without significant administrative overhead. However, it is challenging to make PCA efficient enough for practical use in a low-level system. This talk presents an experimental file system (PCFS), that combines PCA with conditional capabilities to obtain sufficient efficiency, without losing any of PCA's benefits.
(Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.)
The demand for wireless bandwidth in indoor environments such as homes and offices continues to increase rapidly. Although wireless technologies such as MIMO can reach link throughputs of 100s of Mbps (802.11n), the question of how we can deliver 10s of Mbps to a large number of densely-packed nodes remains an open problem. Directional antennas have been shown to be an effective way to increase spatial reuse, but past work has focused largely on outdoor environments where the interactions between wireless links can be ignored. This assumption is not acceptable in dense indoor wireless networks; moreover, indoor deployments need to deal with rich scattering and multi-path effects. In this talk I will introduce DIRC, a wireless network that uses access points with phased array antennas to achieve high throughput in dense, indoor environments. The core of DIRC is an algorithm that identifies the optimal orientations for a network of directional antennas while maximizing overall network capacity. DIRC is implemented and evaluated in a nine-node enterprise setting, which shows that directional antennas can improve overall network capacity in indoor environments, while being flexible enough to adjust to medium variations and the inherent dynamicity of indoor spaces.
(Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.)
