Welcome
The Statistical Inference for Network Models symposium is a satellite of the NetSci2014 conference, to be held June 2, 2014. It will feature a mix of invited and contributed talks, which you can view on the Symposium Schedule. All attendees of this symposium must be registered for the NetSci2014 conference in order to attend. Please note the important dates and deadlines on the right. Symposium Description Network science spans a wide range of scientific disciplines, in which real-world networks are increasingly annotated with rich meta data, including vertex or edge attributes, temporal information, and more. Making sense of such data requires moving beyond simple models of network structure. To understand the complex social, biological, or technological processes that generate these data will require development of new approaches to formulating and testing network hypotheses. Probabilistic modeling offers a principled approach to address these questions. In particular, it allows inference both of structure and structural generating mechanisms while accommodating real data challenges such as noise and missing data, and in doing so, provides interpretability of results, connecting theory to data. Probabilistic methods, in conjunction with computationally efficient forms of inference, are therefore powerful, sophisticated and useful. Yet, they’re not as easy or approachable as degree distribution, centrality, and other classical network analysis tools, and we’re still learning how to use them effectively. There remain important open questions about how to compare hypotheses, and the cost of formulating a new model for a new hypothesis or collecting the correct set of vertex or edge attributes may be high. However, as new applications and richer data sets continue to demand principled techniques capable of answering ever more sophisticated questions, probabilistic methods provide a paradigm for meeting this demand. This satellite symposium aims to unite theoretical and applied researchers by bringing together approaches from machine learning, statistics, physics, and across network science. This broad cross-section of disciplines shares problems and even approaches, but each discipline brings a different perspective, emphasis and vocabulary. The purpose of this symposium is to provide a platform for cross- pollination of ideas and to reveal that the diversity of approaches to a common set of problems is a strength. Invited Speakers - Abstracts and Titles Johan Ugander, Cornell JP Onnela, Harvard Simon Lunagomez, Harvard Tiago Peixoto, Bremen David Choi, Carnegie Mellon Daniel Roy, Cambridge Organizers Dan Larremore, Harvard Abigail Jacobs, Colorado Leto Peel, Colorado Aaron Clauset, Colorado |
Important Dates
|