We consider both finite-horizon and infinite-horizon versions of a dynamic game with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$N$</tex-math></inline-formula> selfish players who observe their types privately take actions that are publicly observed. Players' evolve as conditionally independent Markov processes, conditioned on current actions. Their jointly determine instantaneous rewards. In games...

We study the problem of Bayesian learning in a dynamical system involving strategic agents with asymmetric information. In series seminal papers literature, this has been studied under simplifying model where selfish players appear sequentially and act once game, based on private noisy observations state public observation past players' actions. It is shown that there exist information cascades users discard their mimic action predecessor. paper, we provide framework for studying dynamics...

We consider a finite horizon dynamic game with two players who observe their types privately and take actions, which are publicly observed. Players' evolve as independent, controlled linear Gaussian processes incur quadratic instantaneous costs. This forms (LQG) asymmetric information. show that under certain conditions, players' strategies in private types, together beliefs form perfect Bayesian equilibrium (PBE) of the game. Furthermore, it is shown this signaling due to fact future on...

We consider a finite horizon dynamic game with N selfish players who observe their types privately and take actions, which are publicly observed. Players' evolve as conditionally independent Markov processes, conditioned on current actions. Their actions jointly determine instantaneous rewards. Since each player has different information set, this forms asymmetric there is no known methodology to find perfect Bayesian equilibria (PBE) for such games in general. In paper, we provide two-step...

This paper studies node cooperation in a wireless network from the MAC layer perspective. A simple relay channel with source, relay, and destination is considered where source can transmit packet directly to or through relay. The tradeoff between average energy delay studied by posing problem as stochastic dynamical optimization problem. following two cases are considered: 1) nodes cooperative information decentralized, 2) strategic centralized. With decentralized nodes, structural result...

In this article, we have proposed a model-free reinforcement learning (RL) algorithm, based on sequential decomposition, to obtain optimal policies for mean field games (MFGs). We consider finite horizon MFGs with large population of homogeneous players, sequentially making strategic decisions. Each player observes private state and mean-field representing the empirical distribution other players' states. The is <italic xmlns:mml="http://www.w3.org/1998/Math/MathML"...

In this paper, we consider a finite horizon, non-stationary, mean field game (MFG) with large population of homogeneous players, sequentially making strategic decisions, where each player is affected by other players through an aggregate state termed as state. Each has private type that only it can observe, and representing the empirical distribution players' types, which shared among all them. Recently, authors in [1] provided sequential decomposition algorithm to compute equilibrium (MFE)...

We consider both finite and infinite horizon discounted mean-field games where there is a large population of homogeneous players sequentially making strategic decisions each player affected by other through an aggregate state. Each has private type that only she observes all commonly observe state which represents the empirical distribution players' types. Mean-field equilibrium (MFE) in such defined as solution coupled Bellman dynamic programming backward equation Mckean-Vlasov forward...

The optimal coding scheme for Additive White Gaussian noise (AWGN) channels with noisy output feedback has been unknown several decades. best-known linear is by Chance and Love, where the coefficients of are numerically optimized based on unique observations. In this paper, we introduce a new class schemes, called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">sequential schemes</i> , encoder sequentially updates state process feedback. We...

We consider both finite-horizon and infinite-horizon versions of a dynamic game with N selfish players who observe their types privately take actions that are publicly observed. Players’ evolve as conditionally independent Markov processes, conditioned on current actions. Their jointly determine instantaneous rewards. In games asymmetric information, widely used concept equilibrium is perfect Bayesian (PBE), which consists strategy belief pair simultaneously satisfy sequential rationality...

In this article, we consider a finite-horizon dynamic game with two players who observe their types privately and take actions that are publicly observed. Players' evolve as conditionally independent, controlled linear Gaussian processes, incur quadratic instantaneous costs. This forms asymmetric information. We show under certain conditions, players' strategies in private types, together beliefs, form perfect Bayesian equilibrium (PBE) of the game. is signaling due to fact future beliefs on...

In this paper, we consider a sequential stochastic Stackelberg game with two players, leader, and follower. The follower observes the state of system privately while leader does not. Players play equilibrium where plays best response to leader's strategy. such scenario, has advantage committing policy that maximizes its returns given knowledge is going policy. Such pair strategies both players defined as game. Recently, [1] provided decomposition algorithm compute for games which allow...

In this paper, we consider a discrete time stochastic Stackelberg game where there is defender (also called leader) who has to defend target and an attacker follower). The private type that evolves as controlled Markov process. objective compute Stochastic equilibrium of the commits strategy. attacker's strategy best response defender's optimum given plays response. general computing such involves solving fixed-point equation for whole game. present algorithm computes strategies by smaller...

In this paper, we consider a discrete-time Stackelberg mean field game with leader and an infinite number of followers.The the followers each observe types privately that evolve as conditionally independent controlled Markov processes.The commits to dynamic policy best respond other.Knowing would play based on her policy, chooses maximizes reward.We refer resulting outcome equilibrium (SMFE).In provide master equation allows one compute all SMFE.Based our framework, two numerical...

In this paper, we consider a discrete memoryless point to channel with noisy feedback, where there is sender private message that she wants communicate receiver by sequentially transmitting symbols over channel. After each transmission, receives feedback of the symbol received receiver. The goal design transmission control strategy minimize average probability error. This an instance decentralized information two controllers, and have no common information. There exist methodology in...

In this paper, we consider both finite and infinite horizon discounted dynamic mean-field games where there is a large population of homogeneous players sequentially making strategic decisions each player affected by other through an aggregate state. Each has private type that only she observes.Such have been studied in the literature under simplifying assumption state dynamics are stationary. non-stationary present novel backward recursive algorithm to compute Markov perfect equilibrium...

In this paper, we present a sequential decomposition algorithm to compute graphon mean-field equilibrium (GMFE) of dynamic games (GMFGs). We consider large population players sequentially making strategic decisions where the actions each player affect their neighbors which is captured in graph, generated by known graphon. Each observes private state and also common information as represents empirical networked distribution other players’ types. non-stationary dynamics novel backward...

We consider the problem of how strategic users with asymmetric information can learn an underlying time-varying state in a user-recommendation system. Users who observe private signals about state, sequentially make decision buying product whose value varies time ergodic manner. formulate team as instance decentralized stochastic control and characterize its optimal policies. With users, we design incentives such that reveal their true signals, so gap between objective is small overall...

In this paper, we consider a multi-agent system with N cooperative agents where each agent privately observes its own private type and publicly others' actions. We propose novel model-free reinforcement learning algorithm to compute the optimal policies for that maximizes their collective reward. This setting belongs broad class of decentralized control problems partial information. use common approach [1], wherein some fictitious chooses best policy based on belief current states agents....

Network utility maximization (NUM) is a general framework for designing distributed optimization algorithms networks. Existing studies proposed (economic) mechanisms to solve the NUM but largely neglected issue of large-scale implementation. In this paper, we present Large-Scale Vickery-Clark-Grove (VCG) Mechanism with simpler payment rule. The VCG maximizes network and achieves individual rationality budget balance. We show that, as number agents approaches infinity, each agent's incentive...