Upcoming meetings

On May 23, 3:30-5:00 p.m. in the DSI classroom (Shields Library) we will have our last Network Science Meeting before going on a summer break. We are planning to have 10 minute short talks to showcase the diverse network related research going on at UCD. We encourage all students and postdocs to present their work, especially if you feel that your field has been underrepresented at our meetings. To join the roster, please contact posfai@ucdavis.edu.

As this will be the last event of the season, pizza will be provided at the meeting, please RSVP to help us order enough food for everyone. And after the talks, everyone is invited to join us for an end-of-year celebration at Delta!

Past meetings

December 13th, 2017
Oh-Hyun Kwon -- What Would a Graph Look Like in This Layout? A Machine Learning Approach to Large Graph Visualization
Abstract: Using different methods for laying out a graph can lead to very different visual appearances, with which the viewer perceives different information. Selecting a "good" layout method is thus important for visualizing a graph. The selection can be highly subjective and dependent on the given task. A common approach to selecting a good layout is to use aesthetic criteria and visual inspection. However, fully calculating various layouts and their associated aesthetic metrics is computationally expensive. In this paper, we present a machine learning approach to large graph visualization based on computing the topological similarity of graphs using graph kernels. For a given graph, our approach can show what the graph would look like in different layouts and estimate their corresponding aesthetic metrics. An important contribution of our work is the development of a new framework to design graph kernels. Our experimental study shows that our estimation calculation is considerably faster than computing the actual layouts and their aesthetic metrics. Also, our graph kernels outperform the state-of-the-art ones in both time and accuracy. In addition, we conducted a user study to demonstrate that the topological similarity computed with our graph kernel matches perceptual similarity assessed by human users.

Paper: https://arxiv.org/abs/1710.04328
Preview video: https://vimeo.com/230840405

Novermber 8th, 2017:
Tyler Scott -- Network analysis for messy governance systems
Abstract: Network analysis has gained considerable traction as a tool for understanding the structure and function of complex governance systems. Traditionally, policy network data have stemmed from survey instruments completed by network actors; computational text analysis instead offers the means to use a wide array of procedural texts to generate objective, longitudinal data. This talk demonstrates the use of text mining to observe network relationships. It then shows how these extracted data can be modeled within an inferential network analysis framework, with a particular focus on the conceptual and empirical challenges that arise when studying local governance networks.

May 10th, 2017:
Cuihua (Cindy) Shen and Teresa Gil Lopez -- One Size Fits All: Context Collapse, Self-Presentation Strategies and Language Styles on Facebook
Abstract: This study empirically examines context collapse on Facebook. Context collapse occurs when disparate audiences are conjoined into one audience. Using longitudinal behavioral data of 6,378 users, the study tests how the size and disconnection of people’s social networks are associated with the number of status updates they post and their language variability. Both size and disconnectedness were positively associated with the number of status updates, but negatively associated with language variability. The results suggest that people manage their online self-presentation according to accounts found in the ‘lowest common denominator’ or ‘imagined audience.’ Finally, network size was positively associated with the proportion of positive emotional language and negatively associated with that of negative emotional language, whereas disconnectedness had the opposite effect.

April 19th, 2017:
Nistara Randhawa (with Hugo Mailhot) -- Epidemic spread prediction in data scarce regions
Abstract: While we know how diseases might spread globally via airlines, their local or regional spread is difficult to ascertain in areas with scarce network data. We will present our approach to simulating outbreaks over such regions, utilizing existing GIS information for network construction (e.g. for Rwanda) and subsequent disease (e.g. influenza) spread.

March 8th, 2017:
Hugo Mailhot -- Networked data is more data
Abstract: In this talk, I will present various approaches that have been used to take advantage of the network structure of a given dataset. As a result, we are often able to derive more data from the dataset itself. I'll provide examples in two domains of application:
1. social media user attribute prediction
2. epidemic spreading prediction in data scarce regions

January 11th, 2017:
Keith Burghardt -- How Ebola Spreads: Understanding Past Modeling Mistakes.
Abstract: Recent models of Ebola Virus Disease (EVD) have often made assumptions about how the disease spreads, such as uniform transmissibility and homogeneous mixing within a population. We use the previous outbreak of Ebola as a testbed for these assumptions and offer novel ways to predict and combat the spread of Ebola and similar diseases.
For details: http://www.nature.com/articles/srep34598

Martin Rohden -- Curing critical links in oscillator networks as power flow models
Abstract: Modern societies crucially depend on the robust supply with electric energy so that blackouts of power grids can have far reaching consequences. Typically, large scale blackouts take place after a cascade of failures: The failure of a single infrastructure, such as a critical transmission line, results in several subsequent failures that spread across large parts of the network. Improving the robustness of a network to prevent such secondary failures is thus key for assuring a reliable power supply. We test different strategies to increase transmission capacities to restore stability with respect to transmission line failures. We show that local and nonlocal strategies typically perform alike: One can equally well cure critical links by providing backup capacities locally or by extending the capacities of bottleneck links at remote locations.
For details: https://arxiv.org/abs/1512.00611


June 8th, 2016: Community detection overview and novel optimization based approches
April 27, 2016: Short talks!
March 4, 2016: Chris Smith on multiplexity in organized crime networks and James Sharpnack on denoising over graphs
February 4, 2016: Visualization platform demos
November 30, 2015: ERGMs: A physicist's and social scientist's perspectives
November 2, 2015: Introductions