The proposed research will develop novel methods for analyzing and modeling heterogeneous dynamic networked data. Network data arises in a number of application domains ranging from IoT, cloud computing, software analysis, neuroscience, biology, geography, to social sciences. Accordingly, network analysis has emerged as a major paradigm for exploring complex processes behind observed data. Compared to high dimensional data, analysis of network data is more challenging due to interdependencies between entities, the presence of attributes, and the natural evolution of networks over time.

Multi-armed bandits have been used in multiple contexts that call for a mixture of exploration and exploitation strategies. We are considering this paradigm in the context of repeated assignment of tasks to teams. We assume a latent structured space that encodes the attributes of individuals, teams, and tasks. The goal of the project is to use the structure of the space to develop efficient algorithms for the assignment of single and multiple tasks to teams.

We consider the problem of distinguishing between two groups of subjects based on their white matter connectivity. Our goal is to discriminate between the two groups (the general classification problem in machine learning) not based on global features such as the number of fibers, degree distribution, distribution of streamlines, centrality measures, shortest path distances, but on spatially localized differences in white matter connectivity. Localizing the analysis can reveal important elements that may be lost in global connectivity measures.

Graphs have evolved into a rich framework for the solution of diverse problems, ranging from mathematical puzzles to the analysis of planetary-scale social networks. A particular class of problems that has attracted recent interest in the research community is related to how to model information that is embedded in a graph structure. In this scenario, the graph dynamics is guided by processes that are localized both in time and space, with effects at different scales.

The recent convergence of research in social and psychological sciences, dynamic and quantitative modeling, and network science has led to a re-examination of collective team behavior from a quantitative and systems-oriented viewpoint. Teams cannot be understood fully by studying their components (members) in isolation: team performance is not simply a sum of individual performances; and a diversity of opinions among members leads to better group outcomes. However, it is not yet understood how patterns of interactions and relationships among team members (i.e.

Analysis of opinion dynamics in social networks plays an important role in today's life. For such applications as predicting users' political preference, it is particularly important to be able to analyze the dynamics of competing opinions. Having observed the evolution of polar opinions of a social network's users over time, can we tell when the network "behaved" abnormally? Or, can we predict how the opinions of the users will evolve in future?

Learning a succinct set of substructures that predicts global network properties plays a key role in understanding complex network data. Existing approaches address this problem by sampling the exponential space of all possible subnetworks to find ones of high prediction accuracy. We are develop a novel framework that avoids sampling by formulating the problem of predictive subnetwork learning as node selection, subject to network-constrained regularization.

Past Research Projects

For more information on the lab's previous research projects, please see our publication page.