Ben Y. ZHAO
University of Cambridge, United Kingdom
The emergence of big data requires fundamental new methodology for data analysis, processing, and information extraction. The main challenge here is to perform efficient and robust data processing, while adapting to the underlying resource availability in a dynamic, large-scale computing environment. I would introduce our recent work on the graph processing that have billion-scale of vertices and edges in a commodity single computer, which requires secondary storage as external memory. Executing algorithms results in access to such secondary storage and performance of I/O takes an important role, regardless of the algorithmic complexity or runtime efficiency of the actual algorithm in use.Eiko Yoneki is a Research Fellow in the Systems Research Group of the University of Cambridge Computer Laboratory. She leads a group called ‘data centric systems and networking’, where current research focuses on the exploration of new abstractions for supporting the design and implementation of robust and heterogeneous large-scale data processing.
Efficient large-scale graph processing
The emergence of big data requires fundamental new methodology for data analysis,
processing, and information extraction. The main challenge here is to perform efficient and robust data processing, while adapting to the underlying resource availability in a dynamic, large-scale computing environment. I would introduce our recent work on the graph processing that have billion-scale of vertices and edges in a commodity single computer, which requires secondary storage as external memory. Executing algorithms results in access to such secondary storage and performance of I/O takes an important role, regardless of the algorithmic complexity or runtime efficiency of the actual algorithm in use.
IMT Lucca, Italy
Guido Caldarelli studied Statistical Physics, and he works in the field of Complex Networks. He got his degree in 1992 in Rome (La Sapienza), his PhD in 1996 in Trieste (SISSA). After Postdocs in Manchester and Cambridge he became firstly "Research Assistant" in INFM and secondly "Primo Ricercatore" at ISC-CNR where he is still working with many friends and colleagues. Presently he is Full Professor of Physics at IMT Lucca, and a LIMS Fellow. From November 15th 2015 he is the Vice-President of the Complex Systems Society.
Instability in financial networks
Following the financial crisis of 2007-2008, a deep analogy between the origins of instability in financial systems and in complex ecosystems has been pointed out: in both cases, topological features of network structures influence how easy it is for distress to spread within the system. However, in financial network models, the intricate details of how financial institutions interact typically play a decisive role. Hence, a general understanding of precisely how network topology creates instability remains lacking. Here we show how processes that are widely believed to stabilise the financial system, i.e.market integration and diversification, can actually drive it towards instability, as they contribute to create cyclical structures which tend to amplify financial distress, thereby undermining systemic stability and making large crises more likely. This result holds irrespective of the precise details of how institutions interact, and demonstrates that policy-relevant analysis of the factors affecting financial stability can be carried out while abstracting away from such details.
Ben Y. ZHAO
University California Santa Barbara, USA
Ben Zhao is a Professor at the Computer Science department, U. C. Santa Barbara. He completed his M.S. and Ph.D. degrees in Computer Science at U.C. Berkeley (2000, 2004), and his B.S. from Yale (1997). He is a recipient of the National Science Foundation's CAREER award, MIT Technology Review's TR-35 Award (Young Innovators Under 35), ComputerWorld Magazine's Top 40 Technology Innovators award, Google Faculty awards, the IEEE ITC Early Career Award, and an ACM Distinguished Scientist. His work has been covered by media outlets such as New York Times, Boston Globe, MIT Tech Review, and Slashdot. He has published over 120 publications in areas of security and privacy, networked and distributed systems, wireless networks, data-intensive computing and HCI, with more than 20,000 citations (H-index 51). Finally, he has chaired a number of conferences (WOSN, WWW OSN track, IPTPS, IEEE P2P), and the upcoming World Wide Web Conference (WWW 2016). He is a co-founder and on the steering committee of the ACM Conference on Online Social Networks (COSN).
An empirical view of link prediction in social networks
Algorithms based on complex networks affect our online experience on a daily basis. One of the most ubiquitous examples of this is the link prediction problem, which is a core part of friend recommendations on social networks like LinkedIn, Facebook, and Pinterest, and also part of broader recommendation systems like personal livestreaming on Periscope or Q&A sites like Quora. Given the success of these systems, and the decade of work on link prediction, it is reasonable to assume that this is a solved problem. Yet no quantitative study has been performed to understand just how successful (or unsuccessful) these algorithms are. Meanwhile, there are plenty of anecdotes online of poor recommendations that represent poor prediction results (e.g. Kashmir Hill, Fusion 2016). In this talk, I will present some of our recent work on taking an empirical view to the well studied problem of link prediction in dynamic networks. We implement and apply 18 link prediction algorithms (some metric-based, some machine learning based) to several traces of detailed network dynamics (Renren, Facebook, YouTube), and evaluate their prediction accuracy. We find that on absolute terms, link prediction accuracy is embarrassingly poor across the board, highlighting the fact that this is still very much an open problem. Machine learning approaches tend to outperform relatively, but are often prohibitively high computation costs. We then propose a novel approach to build "prediction filters” using past patterns in network dynamics. Evaluated on our large datasets, our results significantly boost prediction accuracy across all algorithms.
University California Davis, USA
Raissa D’Souza is Professor of Computer Science and of Mechanical Engineering at the University of California, Davis, as well as an External Professor at the Santa Fe Institute. She received a PhD in Statistical Physics from MIT in 1999, then was a postdoctoral fellow at Bell Laboratories and at Microsoft Research. Her interdisciplinary work on network theory spans the fields of statistical physics, theoretical computer science and applied math, and has appeared in journals such as Science, PNAS, and Physical Review Letters. She serves on the editorial board of numerous international mathematics and physics journals, is a member of the World Economic Forum's Global Agenda Council on Complex Systems, and is the President of the Network Science Society.
Steering and controlling systems of interdependent networks
Networks are at the core of modern society, spanning physical, biological and social systems. Each distinct network is typically a complex system, shaped by the collective action of individual agents and displaying emergent behaviors. Moreover, collections of these complex networks often interact and depend upon one another, which can lead to unanticipated consequences such as cascading failures and novel phase transitions. Simple mathematical models of networks can provide important insights into such phenomena. Here we will cover several such models, beginning with control of phase transitions in an individual network then moving on to modeling phenomena in coupled networks, including cascading failures and optimal interdependence.
University of Zaragoza, Spain
Prof. Yamir Moreno is the head of the Complex Systems and Networks Lab (COSNET) since 2003 and is also affiliated to the Department of Theoretical Physics of the Faculty of Sciences, University of Zaragoza. He is the Deputy Director of the Institute for Bio-computation and Physics of Complex Systems (BIFI) and member of its Government Board and Steering Committee. He has been working on nonlinear dynamical systems coupled to complex structures, transport processes and diffusion with applications in communication and technological networks, dynamics of virus and rumors propagation, game theory, systems biology, the study of more complex and realistic scenarios for the modeling of infectious diseases, synchronization phenomena, the emergence of collective behaviors in biological and social environments, the development of new optimization data algorithms and the structure and dynamics of sociotechnicaland biological systems. He has published more than 145 scientific papers in international refereed journals and he serves as reviewer for around 30 scientific journals and research agencies. His research works have collected more than 9300 citations (h=39). At present, he is a member of the Editorial Board of Scientific Reports, Applied Network Science and Journal of Complex Networks, and Academic Editor of PLoS ONE. Prof. Moreno is the elected President of the Complex Systems Society (CSS) and also belongs to its Executive Committee and Council. He is also the Vice-President of the Network Science Society and a member of the Future and Emerging Technology Advisory Group of the European Union’s Research Program: H2020. Besides, he belongs the Advisory Board of the WHO Collaborative Center “Complexity Sciences for Health Systems” (CS4HS), whose headquarters is at the University of British Columbia Centre for Disease Control, in Vancouver, Canada. He is a Fellow of the Institute for Scientific Interchange Foundation (ISI), Turin, Italy since 2013.
On the structure and dynamics of multilayer networks
The availability of highly detailed data of real-world systems have allowed to study systems that are made up of multiple layers. In this talk, we will revise recent advances on the topic of multilayer networked systems. First, we will discuss how to represent these networks and what kind of metrics are needed to capture the new topological complexity arising from the interdependency of the layers. Secondly, we will also study contagion processes on these topologies and present results regarding the interplay between the critical properties of the system and its structural scales. To summarize, we will discuss possible future directions in this new, but very active field of research.
University of Namur, Belgium
After a PhD in Physics at ULB, and Post-docs at ULg, UCLouvain and Imperial College, he is currently associate professor in the Department of Mathematics of the University of Namur. His recent research includes the development of algorithms to uncover information in large-scale networks, the study of empirical data in social and biological systems, and the mathematical modelling of human mobility and diffusion on networks. He has authored more than 60 publications in peer-reviewed journals and conference proceedings, with around 5000 citations (Google Citations). He also acts as an academic editor for PLoS One and the European Physical Journal B.
Burstiness and spreading on networks: Models and predictions
When modelling dynamical systems on networks, it is often assumed that the process is Markovian, that is future states depend only upon the present state and not on the sequence of events that preceded it. Examples include diffusion of ideas or diseases on social networks, or synchronisation of interacting dynamical units. In each case, the dynamics is governed by coupled differential equation, where the coupling is defined by the adjacency matrix of the underlying network. The main purpose of this talk is to challenge this Markovian picture. We will argue that non-Markovian models can provide a more realistic picture in the case of temporal networks where edges change in time, or in situations when pathways can be measured empirically. We will focus on the importance of non-Poisson temporal statistics, and show analytically the impact of burstiness on diffusive dynamics, before turning to applications and incorporating memory kernels in predictive models of retweet dynamics.