Workshop on Dynamics in Social and Economic Networks
59th IEEE Conference on Decision and Control, Jeju Island, Republic of Korea
UTC 1pm - 5 pm, December 12-13, 2020
Zoom meeting link: Event Hall 4, Jeju Virtual Convention Center (JVCC)
*You will need to register for this workshop to obtain an ID and a password for the Zoom meeting.
**For registered attendants: In case of any login issues, please contact Wenjun () or Francesca () for help.
Network effects are pervasive in our society and affect many aspects of our life from how we acquire information, how we interact, how we make decisions, and what opportunities we are exposed to. This is even more the case since the rise of the Internet and online social media, which do not only provide a vast number of empirical data for the quantitative study of social systems, but have also deeply changed the pattern of how people interact with each other. In this era of information revolution and dense interactions, our society faces various unprecedented challenges with profound impacts on modern politics and economy, such as opinion polarization, the politicization of public debates, the effects of echo chambers, and filter bubbles. Phenomena such as spreading of contagion (may this be of pathogen or misinformation), coordination of strategic behavior as well as targeted interventions and incentive for efficient use of resources over networks are rapidly becoming of fundamental importance for both the society and its economy. Mathematical modeling plays a fundamental role in understanding how these macroscopic phenomena emerge from certain microscopic mechanisms of social interactions and certain network structures.
Exploiting the progress in complex networks and data mining, the last decades have witnessed a rapid development of the research on the statistical and static features of social networks, in the framework of Social Network Analysis. However, dynamical processes on/of social networks, which are directly related to the aforementioned phenomena, remain to be thoroughly studied. Due to the rapid progress in the study of multi-agent systems, researchers on control theory have recently contributed various useful mathematical tools to the study and control of social network dynamics. These mathematical tools make it possible for us to investigate some fundamental or emerging problems in social science, including:
(1) What is the "main factor" that governs opinion dynamics and what mechanisms could drive public opinions to polarization?
(2) How do social media and online recommendation system shape the public opinion formation processes?
(3) Is there any efficient way to mitigate the spreading of misinformation or the impact of malicious opinion manipulation?
(4) How does network structure influence individuals' strategic behavior on social networks?
(5) How can one plan targeted interventions or incentives to maximize system performance by exploiting such network effects?
We thus believe this is the perfect timing to bring these socio-economic questions to the attention of the broad audience of control theorists. We organized the workshop by bringing together researchers that work on this rapidly expanding area using different approaches, as for example game theory, complex network analysis, and multi-agent systems so that attendees can have a broad overview of the different techniques that one can use to answer the questions above. The aim is to give a general introduction to dynamics on socio-economic systems, as well as present the latest results on various emerging topics such as social learning and opinion dynamics, network propagation models, coordination of competitive network systems, information design, and interventions under partial information.
Registration for the workshop can be made through this link at the 59th IEEE Conference on Decision and Control website. Please note that only people who have registered for the conference can register for the workshop. The conference early registration rate is till October 1st. The workshop fees are as follows:
Title: Strategic Information Transmission over Networks
Abstract: Strategic information transmission refers to a variation (and a substantial one) of the standard paradigm of information transmission in communication, where now, in its simplest form, the two agents, the sender, and the receiver, have (intentionally) misaligned objectives. This leads to a non-cooperative game with a dynamic (non-classical) information structure, where one can adopt as a game-theoretic solution concept either the Nash or the Stackelberg equilibrium. The talk will introduce this class of problems, which have been of interest to multiple communities, including economics, information theory, communication, signal processing, networking, and control, having picked up considerable steam over the last few years. As an overview of the topic, both old and new results will be presented, including extensions to networks. Strategic information transmission is an important underlying feature of deception games, which will be highlighted in the talk, along with non-trivial extensions to multi-stage multi-agent scenarios, covering sensor networks, cyber-physical systems, and social networks, with also strategic adversarial intervention.
Speaker: Prof. Francesco Bullo, University of California, Santa Barbara
Title: Wisdom of Crowds and Social Influence
Abstract: I will review various models and analysis results for the study of the wisdom of crowds phenomenon and its relationship with social influence. I will present conditions under which an influence system enhances, preserves, or diminishes the wisdom of crowd effect. I will elaborate on the nexus of expertise, self-weight, interpersonal weights, social power and team performance in intellective issues. I will show a fragility property of the wisdom phenomenon, in the presence of irrational individuals.
Speaker: Prof. Jason Marden, University of California, Santa Barbara
Title: The value of information in multiagent coordination
Abstract: The goal in any multiagent system is to derive desirable collective behavior through the design of admissible control algorithms. Here, the admissibility of a given control algorithm is often predicated on the information that is available. This talk will cover three recent results that focus on characterizing how the level of informational availability impacts the achievable performance guarantees in multiagent coordination problems. Our first result focuses on the design of incentive mechanisms to influence and improve societal behavior in congestion games. Here, we will demonstrate how information pertaining to both the infrastructure and population impacts the ability of a societal planner to successfully influence the collective behavior. Our second result shifts attention from influencing societal systems to designing networked engineering systems. Here, we will focus on characterizing how a limited degree of inter-agent communication can be exploited to significantly improve the efficacy of the resulting stable behavior for a class of distributed submodular optimization problems. Lastly, if time permits we will focus on a class of strategic resource allocation games (Colonel Blotto games) and demonstrate how revealing private information can be strategically advantageous in competitive scenarios.
Abstract: Many social and economic systems can be modeled as a network game whereby players are represented as nodes and influences as links of an interaction graph. In this talk, we first discuss some results on the separability structure of network games, refining and generalizing the notion of graphical games. We show that every strategic equivalence class contains a game with minimal separability properties. We prove a symmetry property of the minimal splitting of potential games and we describe how this property reflects to a decomposition of the potential function and generalize this analysis to arbitrary network games showing how their potential-harmonic decomposition relates to their separability properties. We also apply these results in order to evaluate the robustness of Nash equilibria. We then study a targeting problem in network games: the selection of the smallest control set of players capable of driving the system, globally, from one Nash equilibrium to another one. We prove that while the problem is NP-complete even in the special case of the network coordination games. For the class super modular network games, we introduce a randomized algorithm based on a time-reversible Markov chain with provable convergence guarantees.
Prof. Dario Paccagnan, Imperial College London, "Incentivizing efficient use of shared infrastructure: Optimal tolls in congestion games"
Dr. Lorenzo Zino, University of Groningen, "Including cognitive mechanisms in models for social diffusion"
Carlo Cenedese (advisor: Prof. Sergio Grammatico), University of Groningen / TU Delft, topic to be determined
Somya Singh (advisor: Prof. Bahman Gharesifard), Queen's University, "A Finite Memory interacting Polya Contagion Network"
Nicolò Pagan (advisor: Prof. Florian Dörfler), ETH Zurich, "A meritocratic network formation model for the rise of social media influencers"
Bo Liang (advisor: Prof. Lin Wang), Shanghai Jiaotong University, "Structure Preference in Graph Representation Learning Methods"
*Please note that the time in the schedule is the Coordinated Universal Time (UTC), not the Korea local time.
Francesca Parise, Cornell University
University of Groningen
Politecnico di Torino
Slides of talks
by Prof. Jason Marden
by Prof. Giacomo Como
"Analysis of Opinion Dynamics with Bounded and Heterogeneous Confidence Threshold", by Prof. Ge Chen
by Somya Singh
"Asynchronous and Time-Varying Proximal-Type Dynamics" by Carlo Cenedese
"Including Behavioral Mechanisms in Models for Social Diffusion" by Lorenzo Zino
"Online Attention Dynamics: Sociological Modeling and YouTube Case Study" by Maria Castaldo