An the ability to cooperate, coordinate, and

An agent is a computer system that is situated in some environment, and that is capable of autonomous action in this environment
in order to meet its design


systems (MAS) are distributed systems of independent actors, called agents,
that cooperate or compete to achieve a certain objective. These agents may be
computer programs, robots, or even humans. They
can be used to solve problems that are difficult or impossible for an
individual agent or a monolithic system to solve.18 

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now


A cooperative Multi-agent
system (MAS) is composed of a set of autonomous agents that interact with one
another in a shared environment.17 In order to successfully interact, these
agents will thus require the ability to cooperate,
coordinate, and negotiate with
each other, in much the same way that we cooperate, coordinate, and negotiate
with other people in our everyday lives. One
fundamental property of an agent in Multi-agent Systems(MASs) is its ability of
adaptively adjusting its behaviors in response to other agents in order to
achieve effective coordination on desirable outcomes since the outcome not only
depends on the action it takes but also the actions were taken by other agents
that it interacts with. In cooperative MASs, the agents share common interests
(e.g., the same reward function), thus the increase in individual’s benefit
also leads to the increase of the benefits of the whole group. Hao and Leung8
were the first who proposed a multi-agent social learning framework to
investigate multi-agent coordination problem in cooperative games assuming that
the agents’ interactions are random. In their recent work,10 they considered
the underlying network structure instead of random interaction mechanism to
facilitate more efficient co-ordinations. To this end, in this work, we study
the previous works that have been done to improve the coordination in a
cooperative multi-agent system under the multi-agent learning frameworks. We
shall introduce a proposed networked multi-agent learning framework to investigate
the multi-agent coordination problem and to improve multi-agent coordination
efficiency in cooperative MASs by explicitly modeling different network
topologies and by considering some characteristics of different neighboring
agents. In this framework, each agent learns its policy through repeated
interactions with its neighboring agents in the system. We shall consider a
number of representative social network structures: ring network, small-world
network and scale-free network. During each round, each agent interacts with
one of its neighbors randomly, and the interactions between each pair of agents
are modeled as two-player cooperative Markov games. If no underlying topology
exists, then one agent is randomly selected as its partner from the population.
Each agent learns its policy concurrently over repeated interactions with
randomly selected partners from its neighborhood. Besides, apart from learning
from its own experience, each agent may also learn from the experience of its
neighbors. We believe that more work needs to be done on top of all the
previous frameworks to improve the coordination efficiency in different
scenarios. Therefore, we
are utilizing the characteristics of different neighboring agents (e.g., the
relative degrees of different nodes, the past coordination histories of
different nodes) when performing the multi-agent learning (the observation
mechanism) to improve the coordination performance. This is the unique
characteristic of network-based multi-agent social learning which is expected
to be able to facilitate the coordination of agents on optimal outcomes.