Luis BETTENCOURT - University of Chicago, USA
Michelle GIRVAN - University of Maryland , USA
Kimberly GLASS - Harvard Medical School Boston, USA
Mark NEWMAN - University of Michigan, USA
David STARK - Columbia University New York, USA
Complex systems are the result of diverse interactions, which can often be modeled by networks or graphs. It is the virtue of graphs that they simplify physical relationships and forces, but such abstractions can also leave important problems under-constrained and hard to solve fundamentally. Social life in cities provides a nice environment where networks become a crucial part of mathematical modeling, but where the physical and social embeddings of links can also be understood and derived across scales. In this talk, I will introduce networks as a tool to understand two types of interdependent structures in cities: socioeconomic relationships between people, and networks of infrastructure between places. I will show that both of these networks have specific general properties that are self-similar with city size, allowing us to predict the properties of hitherto unstudied human settlements in history and in the future. I will finish by showing that the growth of these networks is the measurable physical manifestation of the process of development in societies, opening up new opportunities for interdisciplinary theory and for more effective policy.
Biography
Luís M. A. Bettencourt is a Professor of Ecology and Evolution and the College at the University of Chicago. He is also Associate Faculty of the Department of Sociology and External Professor at the Santa Fe Institute. He grew up in Lisbon (Portugal) and obtained his undergraduate degree in Engineering Physics from IST Lisbon. He obtained his PhD from Imperial College London in Theoretical Physics and held postdocs and research positions at the University of Heidelberg (Germany), Los Alamos National Laboratory, MIT, and the Santa Fe Institute. His research focuses on the theory and modeling of complex systems and the processes that underlie the structure and growth of cities, in particular. He connects interdisciplinary concepts and advanced mathematicswith new technologies and data to create new systems’ theory and methods. This work also involves collaborations with governments, NGOs, and interdisciplinary researchers worldwide to co-produce new insights and transformative practices for sustainable development. His work is well-known academically and widely covered in the media. It has helped shape our fundamental understanding of complex systems and human societies and create novel approaches to challenges of urbanization and sustainability.
Federico Battiston, an Associate Professor in Network Science, brings a wealth of experience from esteemed institutions like CEU, University College London, and the Brain & Spine Institute in Paris. Holding a PhD from Queen Mary University of London, his expertise spans statistical physics, complexity science, and social networks. As Chair of NetSci2023, he oversees the premier conference in network science, demonstrating his leadership in the field. Federico's research, published in prestigious journals like Nature Physics and Science Advances, delves into diverse topics, from sustainable urban systems to the human brain. Recognized with awards such as the Junior Award of the Complex Systems Society and the Early Career Prize in Statistical and Nonlinear Physics of the European Physical Society, Federico continues to advance our understanding of complex systems and network dynamics, mentoring a new generation of researchers.
Understanding and forecasting the behavior of complex dynamical systems is a central challenge in network science, especially when data are sparse, heterogeneous, or drawn from multiple interacting subsystems. In this talk, I will highlight some of my group’s work at the intersection of machine learning, nonlinear dynamics, and complex networks, with a focus on reservoir computing as a powerful and flexible framework for learning system dynamics from time-series data. In this context, network structure plays a dual role: first, the reservoir itself is a randomly structured recurrent neural network whose inherent complexity gives rise to the rich internal dynamics that make the approach effective; second, the framework is naturally suited for predicting dynamics on networks—such as disease spread, synchronization, or activation patterns—making it especially relevant to the network science community. In the second part of the talk, I will introduce our recent method, Meta-learning for Tailored Forecasting from Related Time Series (MeTaFoRS), which addresses the challenge of forecasting from short and heterogeneous time series by drawing on a library of models trained on related systems. Using reservoir computing and testing on families of simulated chaotic systems, we demonstrate that MeTaFoRS can recover both short-term trajectories and long-term statistics, even when test and training systems exhibit markedly different behaviors, highlighting the method's robustness and versatility in data-limited scenarios.
Biography
Michelle Girvan received her B.S. in 1999 from the Massachusetts Institute of Technology and her Ph.D. in 2003 from Cornell University. Her research combines methods from statistical mechanics, dynamical systems, and graph theory to address interdisciplinary, network-related problems. She is interested in both broad theoretical approaches to complex networks as well as specific applications, especially to information cascades, epidemiology, and genetic regulatory networks. In a 2019 podcast, she discussed her work in chaos and artificial intelligence. In 2022, she was named a UMD Distinguished Scholar-Teacher.
Tina Eliassi-Rad is the inaugural President Joseph E. Aoun Professor at Northeastern University. She is also an external faculty member at the Santa Fe Institute and the Vermont Complex Systems Center. Tina works at the intersection of artificial intelligence and network science and is interested in the impact of science and technology on society. For a more extended bio, visit http://eliassi.org/bio.html .
Combining biomedical data with networks can provide unprecedented insights into the mechanisms underlying disease. Furthermore, quantifying how the complex structure of biological networks is altered during disease is critical for developing new therapeutic or prevention strategies. Although many methods have been developed to estimate biological networks, these approaches typically use multiple experimental samples to estimate a single “aggregate” network and fail to capture population-level heterogeneity. In this talk I will review several approaches my group has developed for inferring networks from biomedical data, including a mathematical framework for estimating sample-specific networks. I will show how these approaches can be used to associate complex network connectivity patterns with other sample-specific information, such as patient phenotype. Finally, I will examine the caveats and underlying assumptions made by these methods, highlighting some of the broader challenges and opportunities in the emerging field of precision network medicine.
Biography
Kimberly Glass is an expert in complex networks and genomic data analysis. She obtained her PhD in Physics in 2010 from the University of Maryland. From 2010-2014, Dr. Glass was a postdoctoral fellow at Dana-Farber Cancer Institute and the Harvard T.H. Chan School of Public Health where she received training in computational biology. During her post-doc she developed several computational and data-integration methods for inferring and analyzing gene regulatory networks. In 2014 Kimberly joined the faculty of the Channing Division of Network Medicine (CDNM) at Brigham and Women’s Hospital where she is continuing her research in systems medicine and network methods. Her current research focuses on how to integrate and interpret multiple biological data-types in the regulatory network context and on how to understand the biological mechanisms represented in these networks. She is also investigating potential applications of networks in precision medicine, using network approaches to understand susceptibility to, severity, and treatment of complex diseases.
Frank Emmert-Streib, a distinguished Professor of Data Science at Tampere University, leads the Predictive Society and Data Analytics Lab, focusing on interdisciplinary research in data science. Formerly a Senior Lecturer at Queen's University Belfast, his expertise spans biostatistics, computational biology, and theoretical physics. Frank received training as a Senior Fellow at the University of Washington and was a Postdoctoral Research Associate at the Stowers Institute for Medical Research. With a Ph.D. in Theoretical Physics from the University of Bremen, he has significantly contributed to computational and statistical methods, particularly in addressing uncertainty and explainability in data analysis. Frank's extensive experience includes sabbaticals at prestigious institutions like Harvard School of Public Health and the University of Cambridge. As a co-founder and former CSO of sAnalytiCO Ltd, he has demonstrated leadership in academia and industry, shaping the future of data technology. He plays an active role in academic publishing as an editor and associate editor for prestigious scientific journals.
One of the oldest of network problems is the ranking of individuals, teams, or commodities on the basis of pairwise comparisons between them. For example, if you know which football teams beat which others in a particular year, can you say which team is the best overall? This is a harder problem than it sounds because not all pairs of teams play one another in a given season, and also because the outcomes of the games can be contradictory. This talk will introduce the techniques used to solve such ranking problems, with examples from games and sports, consumer research and marketing, and social hierarchies in both animal and human communities, then ask how those techniques can be extended to address a range of new questions about competition and ranking, including the development of new computer algorithms for ranking, questions about the varying patterns of competition in different sports, and what happens when individuals or teams compete in multiple different ways.
Biography
Professor Newman's research is on statistical physics and the theory of complex systems, with a primary focus on networked systems, including social, biological, and computer networks, which are studied using a combination of empirical methods, analysis, and computer simulation. Among other topics, he and his collaborators have worked on mathematical models of network structure, computer algorithms for analyzing network data, and applications of network theory to a wide variety of specific problems, including the spread of disease through human populations and the spread of computer viruses among computers, the patterns of collaboration of scientists and business-people, citation networks of scientific articles and law cases, network navigation algorithms and the design of distributed databases, and the robustness of networks to the failure of their nodes. Professor Newman also has a research interest in cartography and was, along with collaborators, one of the developers of a new type of map projection or "cartogram" that can be used to represent geographic data by varying the sizes of states, countries, or regions. Professor Newman is the author of several books, including a recent textbook on network theory and a popular book of cartography.
Filippo Menczer is the Luddy distinguished professor of informatics and computer science and the director of the Observatory on Social Media at Indiana University. He holds a Laurea in Physics from the Sapienza University of Rome and a Ph.D. in Computer Science and Cognitive Science from the University of California, San Diego. His research interests span Web and data science, computational social science, science of science, and modeling of complex information networks. Dr. Menczer was named a Fellow of the ACM for his research on the vulnerability of social media networks to disinformation and manipulation.
In many diverse settings, preparations for vital systems security are tested by various kinds of experiments, scenarios, simulations, and enactments. But recent theorization and research about preparedness is changing the problem: from the concept of preparing for a broad range of imagined futures, preparedness is now about resilience in the face of unanticipated, indeed unanticipatable, calamity. In this problematic, what happens to the notion of testing?
Whereas preparedness might be thought about as pattern recognition (be prepared to take action x in situation y), adaptation in the face of extraordinary uncertain is about re-cognition – seeing possibilities that were not already imaginable. This same type of recognition is characteristic, indeed defining, of innovation: “Innovation is not having new resources to accomplish new tasks but recognizing configurations that others would not see as resources” (Beunza and Stark 2003: 148).
For most people writing in the field of preparedness, the problem to be anticipated, mitigated, or governed is a potentially calamitous, negative event. What if, instead, we thought about an organization being prepared for something positive? For organizational theorists, that effectively translates to questions about what prepares an organization for innovation. What are the structures and practices, but also the conditions, that make it more likely that an organization will create surprises?
Biography
David Stark is Arthur Lehman Professor of Sociology at Columbia University where he directs the Center on Organizational Innovation. A major figure in the field of economic sociology, he uses a broad variety of research methods – ethnographic, network analytic, and experimental – to study processes of valuation and innovation. He has studied factory workers in socialist Hungary, new media employees in a Silicon Alley startup, derivative traders on Wall Street, electronic music artists in Berlin, bankers in Budapest, farmers in Nebraska, video game producers, and megachurches that look like shopping malls.
In his book, The Sense of Dissonance: Accounts of Worth in Economic Life (Princeton University Press, 2011), Stark shows how organizations and their members search for what is valuable. Dissonance – disagreement about the principles of worth – can lead to discovery. Postsocialist Pathways (with Laszlo Bruszt, Cambridge U Press, 1998) compares the different ways in which the societies of East Central Europe dealt with the challenge of simultaneously transforming property rights and citizenship rights. In an oft-cited article, “Recombinant Property in East European Capitalism,” (American Journal of Sociology 1996) he argues that capitalism in Eastern Europe was not built on the ruins of communism but with the ruins of communism.
Stark recently completed a major research project on Diversity and Performance: Networks of Cognition in Markets and Teams supported by a five-year Advanced career Award from the European Research Council. Recent publications include “Racial Attention Deficit” (Science Advances 2021), “Put to the Test: For a New Sociology of Testing” (British Journal of Sociology 2020), and The Performance Complex: Competition and Competitions in Social Life (Oxford University Press 2020). His current research project: Algorithmic Management and New Class Conflicts.
He was named a Guggenheim Fellow in 2002 and was awarded an honorary doctorate from the École normale supérieure de Cachan in 2013. Stark has been a visiting fellow at numerous institutes including: Wissenschaftskolleg, Berlin; Center for Advanced Study in the Behavioral Sciences, Palo Alto; Max Planck Institute for the Study of Societies, Cologne; the Russell Sage Foundation; Collegium Budapest; and the Institute for Advanced Study in Hangzhou, China.
Stark’s edited collection, Practicing Sociology: Tacit Knowledge for the Sociological Craft, is in press at Columbia University Press. Many of his articles, books, presentations, “silent lectures,” and other materials are available at https://davidcstark.net.
Luciano Pietronero graduated in Physics in 1971 in Rome. After several experiences abroad in the corporate research sector (Xerox Research Center Webster in New York and Brown Boveri Research Center in Switzerland), he was a full professor of Condensed Matter Physics at the University of Groningen in the Netherlands and then at the University of Rome “Sapienza”. In 2004, he founded the Institute of Complex Systems (ISC) of the CNR. In 2007, he was the president of the 23rd edition of the STATPHYS conference, and in 2008, he won the Enrico Fermi prize, the most important of the Italian Physics Society. Author or co-author of more than 400 scientific publications; in 1987, he introduced the concept of fractal cosmology and in 2012, the Economic Fitness and Complexity model. Mentor of a generation of young scientists in the fields of complex systems, statistical mechanics and superconductivity at high temperatures, since 2019, he has been President of the Enrico Fermi Historical Museum of Physics and Study and Research Centre.