Spring 06 Informal Seminar Events

Ted Senator and Michael Kearns

Possible Experiments on Collective Problem Solving

TBA

This week there will be no formal speaker. Instead we will be discussing as a group serveral papers that examine the role of travel in understanding global epidemics. The two main papers are: 1) Vittoria Colizza, Alain Barrat, Marc Barthelemy, and Alessandro Vespignani "Prediction and predictability of global epidemics:the role of the airline transportation network" Complexity Digest 2005.32 (2005) and 2) D. Brockmann, L. Hufnagel, and T. Geisel "The scaling laws of human travel" Nature 439 (2006). Additionally, here are two related papers: 1) Rory Howlett "Tavel: Fitting the Bill" Nature 439 (2006) (http://www.nature.com/nature/journal/v439/n7075/full/439402b.html) and 2) Duncan J. Watts, Roby Muhamad, Daniel C. Medina, and Peter S. Dodds "Multiscale, resurgent epidemics in a hierarchical metapopulation model" PNAS 102:32 (2005).

[colizza2006airlinesAndEpidemics.pdf] [brockmann2006_humanTravel.pdf] [watts2005_epidemics.pdf]

Chris Wiggins
Assistant Professor, Applied Physics and Applied Mathematics, Columbia

Predicting Evolutionary Design From Network Topology: A Machine Learning Approach

The past 5 years have seen a resurgence of interest in network modeling, particularly in competing models of networks that grow (in the case of technological or social networks) or evolve (in the case of biological networks). While predominant analytic approaches focus on degree distribution or other statistics that offer limited information about the underlying evolutionary mechanisms, a machine learning approach can clearly, quantitatively, and interpretably distinguish among competing explanations for the design principles (of lack thereof) giving rise to the topology of biological networks. We study the protein-protein interaction network of D. melanogasterin in detail but apply the approach to a number of other biological networks. The approach developed can be seen as a special case of a more general and understudied application of classification: as a form of model selection for finding the best candidate among competing generative models attempting to model rich, structured data, such as the networks commonly studied in systems biology

[middendorf2004_biologicalNetwork.pdf] [middendorf2004_dmelanogaster.pdf]

Nassim Nicholas Taleb
Dean's Professor, Sciences of Uncertainty, UMASS at Amherst

Mild vs. Wild Randomness

The talk is on the nontrivial difference between Mild (Gaussian) and Wild randomness (non-Gaussian), its consequences for knowledge, prediction, and social fairness, and how it renders much of the statistical tools ineffectual. Related papers can be found on Taleb's website: http://www.fooledbyrandomness.com/. Of particular relevance are /epistemologyfattails.pdf, /knolwedge.pdf, and /amherstclass/blackswanexcerpts.pdf (the last of which requires a username and password which will be included in the email announcement).

Gueorgi Kossinets
Graduate Student, Sociology, Columbia

A Discussion on Response Times Derived from Email Logs

As opposed to a formal lecture, Geuorgi will be leading a discussion, so please familiarize yourself with the following two papers before the talk: (1) "The origin of bursts and heavy tails in human dynamics" by Albert-László Barabási [http://www.nature.com/nature/journal/v435/n7039/full/nature03459.html] and (2) a longer version of the first paper: "Modeling bursts and heavy tails in human dynamics" by A. Vazquez et al. [http://arxiv.org/abs/physics/0510117].

Simon Levin
Professor, Ecology & Biology, Princeton

The emergence of collective decision-making

I will explore models of collective behavior and collective decision-making in non-humans, and speculate on implications for human collectives

Peter Hoff
Assistant Professor, Statistics, University of Washington

Latent factor models for relational data

atrix representation techniques have a long history in the analysis of multivariate data, including relational data in which observations are on pairs of individuals or units. In particular, the singular value decomposition of a matrix allows one to represent the relationship between two units as the inner product of a pair of latent characteristic vectors. In this talk I discuss a model-based version of the singular value decomposition which allows for the analysis of a variety of data types, including binary relational data, or social networks. Two recent papers on the subject can be found at: www.stat.washington.edu/hoff/Preprints/lfm.pdf and www.stat.washington.edu/hoff/Preprints/tr494.pdf

Peter Hedstrom
Professor, Sociology, Nuffield College at the University of Oxford

Tie-Formation Mechanisms and the Evolution of Labor-Market Networks

Orkut Buyukkokten
Software Engineer, Google

Online Social Networks

Online social networks fundamentally change the way we get connected. The people we cross paths with have the biggest influence in our lives. Now it's easer to cross paths than ever as we are much closer and so much more connected. The possibilities are endless. In this talk I will discuss the motivation behind the development of orkut.com, touch on the technical aspect of implementing and maintaining a system that has over 15 million users and reflect on the lessons learned from observing the growth.

H. Peyton Young
Professor, Economics, Johns Hopkins

The Spread of Innovations by Social Learning

One common explanation for the diffusion of innovations is information contagion: agents adopt once they hear about the existence of the innovation from someone else. An alternative explanation is learning: agents adopt once the perceived gain from adoption -- as revealed by the outcomes among prior adopters -- exceeds a threshold determined by the agents’ prior beliefs. We demonstrate that learning with heterogeneous beliefs generates an adoption dynamic that has a simple and completely general closed-form solution. We also show that the pattern of acceleration generated by learning often differs qualitatively from the pattern generated by contagion, even when both give rise to S-curves. In particular, learning may produce super-exponential growth rates in the early stages of adoption, whereas standard contagion models do not have this property. These super-exponential growth rates are observed in Griliches’ landmark study of the diffusion of hybrid corn.

[young2006_innovation.pdf]

Alan Kirman
Member, School of Social Science, Institute for Advanced Study

Chasing Identity: The Emergence of Social Groups

Ling Bian
Associate Professor, Geography, University of Buffalo

A Conceptual Framework for an Individual-Based Spatially Explicit Epidemiological Model

The talk will be from 2-4 PM (talk is 1 hour, she’ll be around for questions for another hour). This is a joint venture between the RWJ Infectious Disease Working Group and the Collective Dynamics Group.

Anna Goldenberg
Graduate Student, Computer Science, Carnegie Mellon

Contextual Dynamic Friendship Networks

The study of social networks has gained new importance with the recent rise of large on-line communities. Most current approaches focus on deterministic (descriptive) models and are usually restricted to a preset number of social actors. Moreover, the dynamic aspect is often treated as an addendum to the static model. Taking inspiration from real-life friendship formation patterns, we propose a new generative model of evolving social networks that allows for birth and death of social ties and addition of new actors. Each actor has a distribution over social interaction spheres, which we term "contexts." We study the robustness of our model by examining statistical properties of simulated networks relative to well known properties of real social networks. A Gibbs sampling procedure is developed for parameter learning.

[goldenberg2006_friendship.pdf]