arXiv subjects

Ceyhun Eksin

Publications and source records attributed to Ceyhun Eksin.

26 records · Page 2Linked to original sources

arXiv2018

Control of learning in anti-coordination network games

We consider control of heterogeneous players repeatedly playing an anti-coordination network game. In an anti-coordination game, each player has an incentive to differentiate its action from its neighbors. At each round of play, players take actions according to a learning algorithm that mimics the iterated elimination of strictly dominated strategies. We show that the learning dynamics may fail to reach anti-coordination in certain scenarios. We formulate an optimization problem with the objective to reach maximum anti-coordination while minimizing the number of players to control. We consider both static and dynamic control policy formulations. Relating the problem to a minimum vertex cover problem on bipartite networks, we develop a feasible dynamic policy that is efficient to compute. Solving for optimal policies on benchmark networks show that the vertex cover based policy can be a loose upper bound when there is a potential to make use of cascades caused by the learning dynamics of uncontrolled players. We propose an algorithm that finds feasible, though possibly suboptimal, policies by sequentially adding players to control considering their cascade potential. Numerical experiments on random networks show the cascade-based algorithm can lower the control effort significantly compared to simpler control schemes.

eess.SY↗

arXiv2018

Distributed Inertial Best-Response Dynamics

The note considers the problem of computing pure Nash equilibrium (NE) strategies in distributed (i.e., network-based) settings. The paper studies a class of inertial best response dynamics based on the fictitious play (FP) algorithm. It is shown that inertial best response dynamics are robust to informational limitations common in distributed settings. Fully distributed variants of FP with inertia and joint strategy FP with inertia are developed and convergence is proven to the set of pure NE. The distributed algorithms rely on consensus methods. Results are validated using numerical simulations.

eess.SY↗

arXiv2018

Optimal control policies for evolutionary dynamics with environmental feedback

We study a dynamical model of a population of cooperators and defectors whose actions have long-term consequences on environmental "commons" - what we term the "resource". Cooperators contribute to restoring the resource whereas defectors degrade it. The population dynamics evolve according to a replicator equation coupled with an environmental state. Our goal is to identify methods of influencing the population with the objective to maximize accumulation of the resource. In particular, we consider strategies that modify individual-level incentives. We then extend the model to incorporate a public opinion state that imperfectly tracks the true environmental state, and study strategies that influence opinion. We formulate optimal control problems and solve them using numerical techniques to characterize locally optimal control policies for three problem formulations: 1) control of incentives, and control of opinions through 2) propaganda-like strategies and 3) awareness campaigns. We show numerically that the resulting controllers in all formulations achieve the objective, albeit with an unintended consequence. The resulting dynamics include cycles between low and high resource states - a dynamical regime termed an "oscillating tragedy of the commons". This outcome may have desirable average properties, but includes risks to resource depletion. Our findings suggest the need for new approaches to controlling coupled population-environment dynamics.

eess.SY↗

arXiv2016

Demand Response with Communicating Rational Consumers

The performance of an energy system under a real-time pricing mechanism depends on the consumption behavior of its customers, which involves uncertainties. In this paper, we consider a system operator that charges its customers with a real-time price that depends on the total realized consumption. Customers have unknown and heterogeneous consumption preferences. We propose behavior models in which customers act selfishly, altruistically or as welfare-maximizers. In addition, we consider information models where customers keep their consumption levels private, communicate with a neighboring set of customers, or receive broadcasted demand from the operator. Our analysis focuses on the dispersion of the system performance under different consumption models. To this end, for each pair of behavior and information model we define and characterize optimal rational behavior, and provide a local algorithm that can be implemented by the consumption scheduler devices. Analytical comparisons of the two extreme information models, namely, private and complete information models, show that communication model reduces demand uncertainty while having negligible effect on aggregate consumer utility and welfare. In addition, we show the impact of real-time price policy parameters have on the expected welfare loss due to selfish behavior affording critical policy insights.

eess.SY↗

arXiv2016

Networked SIS Epidemics with Awareness

We study an SIS epidemic process over a static contact network where the nodes have partial information about the epidemic state. They react by limiting their interactions with their neighbors when they believe the epidemic is currently prevalent. A node's awareness is weighted by the fraction of infected neighbors in their social network, and a global broadcast of the fraction of infected nodes in the entire network. The dynamics of the benchmark (no awareness) and awareness models are described by discrete-time Markov chains, from which mean-field approximations (MFA) are derived. The states of the MFA are interpreted as the nodes' probabilities of being infected. We show a sufficient condition for existence of a "metastable", or endemic, state of the awareness model coincides with that of the benchmark model. Furthermore, we use a coupling technique to give a full stochastic comparison analysis between the two chains, which serves as a probabilistic analogue to the MFA analysis. In particular, we show that adding awareness reduces the expectation of any epidemic metric on the space of sample paths, e.g. eradication time or total infections. We characterize the reduction in expectations in terms of the coupling distribution. In simulations, we evaluate the effect social distancing has on contact networks from different random graph families (geometric, Erdős-Renyi, and scale-free random networks).

cs.SI↗

arXiv2016

Disease dynamics on a network game: a little empathy goes a long way

Individuals change their behavior during an epidemic in response to whether they and/or those they interact with are healthy or sick. Healthy individuals are concerned about contracting a disease from their sick contacts and may utilize protective measures. Sick individuals may be concerned with spreading the disease to their healthy contacts and adopt preemptive measures. Yet, in practice both protective and preemptive changes in behavior come with costs. This paper proposes a stochastic network disease game model that captures the self-interests of individuals during the spread of a susceptible-infected-susceptible (SIS) disease where individuals react to current risk of disease spread, and their reactions together with the current state of the disease stochastically determine the next stage of the disease. We show that there is a critical level of concern, i.e., empathy, by the sick individuals above which disease is eradicated fast. Furthermore, we find that if the network and disease parameters are above the epidemic threshold, the risk averse behavior by the healthy individuals cannot eradicate the disease without the preemptive measures of the sick individuals. This imbalance in the role played by the response of the infected versus the susceptible individuals in disease eradication affords critical policy insights.

cs.SI↗

arXiv2016

Distributed Fictitious Play for Optimal Behavior of Multi-Agent Systems with Incomplete Information

A multi-agent system operates in an uncertain environment about which agents have different and time varying beliefs that, as time progresses, converge to a common belief. A global utility function that depends on the realized state of the environment and actions of all the agents determines the system's optimal behavior. We define the asymptotically optimal action profile as an equilibrium of the potential game defined by considering the expected utility with respect to the asymptotic belief. At finite time, however, agents have not entirely congruous beliefs about the state of the environment and may select conflicting actions. This paper proposes a variation of the fictitious play algorithm which is proven to converge to equilibrium actions if the state beliefs converge to a common distribution at a rate that is at least linear. In conventional fictitious play, agents build beliefs on others' future behavior by computing histograms of past actions and best respond to their expected payoffs integrated with respect to these histograms. In the variations developed here histograms are built using knowledge of actions taken by nearby nodes and best responses are further integrated with respect to the local beliefs on the state of the environment. We exemplify the use of the algorithm in coordination and target covering games.

cs.GT↗

arXiv2013

Bayesian Quadratic Network Game Filters

A repeated network game where agents have quadratic utilities that depend on information externalities -- an unknown underlying state -- as well as payoff externalities -- the actions of all other agents in the network -- is considered. Agents play Bayesian Nash Equilibrium strategies with respect to their beliefs on the state of the world and the actions of all other nodes in the network. These beliefs are refined over subsequent stages based on the observed actions of neighboring peers. This paper introduces the Quadratic Network Game (QNG) filter that agents can run locally to update their beliefs, select corresponding optimal actions, and eventually learn a sufficient statistic of the network's state. The QNG filter is demonstrated on a Cournot market competition game and a coordination game to implement navigation of an autonomous team.

eess.SY↗