In this paper we develop and estimate a simple multi-agent model for the U.S. housing mar- ket. The development of better models for the housing market is of high importance, in the light of the failure of financial institutions and regulators to predict the house price collapse that triggered the global financial crisis in 2008–2009. The housing market is more vulnerable to inefficiencies than other markets due to lack of effective short selling mechanisms that pre- vent bearish investors from participating. Furthermore, the heterogeneity in housing stock as well as the heterogeneity in market participants prevents standard arbitrage processes from functioning properly. Our main result is that the interaction between agents in the model can generate boom-bust cycles endogenously, even in the absence of underlying fundamental news. Agents in the model can switch between the fundamentalist and chartist forecasting rules, depending on the rules' recent prediction performance. Precisely this feature allows the market to be driven by chartists when a price bubble builds up, but dominated by fundamentalists dur- ing the eventual burst [4,5]. In the companion article , we further show that the econometric model derived from this multi-agentsystem delivers better out-of-sample price forecasts for the U.S. housing market than standard models.
As we stated in the introduction of our paper, the problems related to airline scheduling and operations are, traditionally, solved using Operations Research techniques. We believe that, using MAS technology, we can go a step forward and create a more autonomous system without sacrifying the demanding commercial and operational requirements of this domain. What distinguishes our approach from Operations Research traditional approaches is the wider scope of our MAS. We want the users to be managers and not controllers. In our MAS the decision support tools of the OR techniques will be agents that will compete and/or collaborate to find the best solution to the problems that may appear during airline operations control. Finally, our MAS will take advantage of the geographic distribution of resources through operational bases. As a summary, we are developing a Distributed Multi-agentSystem, with emphasis in the following:
Rockel et al.  present a method to integrate Jadex multi-agentsystem  with a multi-robot system Player/Stage  by implementing an intermediate layer between them. The intermediate layer embeds data, robot, device and behavior components and manages the transition between a synchronous interface to the robot hardware and an asynchronous interface provided to the MAS. The data component contains central data types used throughout other components. The Robot component contains a generic Robot class from which implemented specialized robots inherit. From the specialized Robot component, Device components are organized according to the robot architecture to interface with Player/Stage client to have mechanical actions properly executed by device drivers. The Behavior component implements a basic set of behaviors for a mobile robot through communication services provided by the MAS and which are used to create information channels to which an agent can subscribe in order to read or publish interesting information to services according to its activities and abilities.
From previous works, it was noted that function parallelism was used to shorten the delay in the processing of the packets; however, it did not reduce the delay of signature matching in each node due to the use of multi- pattern search algorithms. Therefore, distributing the rules across nodes only lessened the processing delay. In contrast, data parallelism reduced the inspection time in each node due to the use of bounded amounts of data; thus, decreasing the overall processing time of the packets. All the approaches presented with data parallelism were aimed at constructing a misuse detection system, without any mention of approaching aims to build an anomaly detection system; in which such approaches depend upon matching the attack signatures with incoming network traffic. In addition, all previous researches interested with IDSs that employed parallelism in their work to reduce the time required for processing had applied various methods to construct the parallel system; but so far, no research has utilised the Multi-agentSystem as a distribution system to create parallelism for reducing the time required for processing data packets in IDS.
chain typically extends across the multiple enterprises including suppliers, manufacturers, transportation carriers, ware houses, retailers as well as customers and entails sharing forecast, order, inventory, and production information to better coordinate management decisions at multiple points throughout the extended enterprise. The basic characteristics of the supply chain indicate that the efforts associated with the formation of supply chain can be easily extended to the formation of virtual enterprise. Hence, the proposed work intends to focus on the realization of virtual enterprise through the formation of supply chain as the initial step. Agents have revolutionized manufacturing systems. Agent technology provides a natural way to design and implement distributed intelligent manufacturing environments and provides software architecture for managing the supply chain. In distributed intelligent manufacturing systems, the main function of agents is to integrate manufacturing enterprise activities such as design, planning, execution, simulation, distribution, forecasting between suppliers, customers and partners. They are also used to represent various manufacturing sources like products, parts and operations to facilitate different manufacturing activities. With this idea a research activity has been identified to develop a multiagentsystem model for designing supply chain which result in efficient sharing of information and integrated functioning of various units of an organization and also enhances the communication with other collaborating enterprises
Fig. 5 shows the proposed multi-agentsystem called MAgSeM. The dotted lines show instantiated agents. At the sender’s side, IA sends the ID, password and IP address of the sender to SvA to be authenticated (assumption is made that the certificates of all users have been exchanged beforehand. A security administrator in an organization, such as a hospital could be responsible for managing certification exchanges). SvA authenticates the user and if the sender is authorized, it sends the authentication result (valid/invalid) as well as a list of IP addresses of other users that have exchanged certificates with the sender.
Abstract: An airline schedule very rarely operates as planned. Problems related with aircrafts, crew members and passengers are common and the actions towards the solution of these problems are usually known as operations recovery or disruption management. The Airline Operations Control Center (AOCC) tries to solve these problems with the minimum impact in the airline schedule, with the minimum cost and, at the same time, satisfying all the required safety rules. Usually, each problem is treated separately and some tools have been proposed to help in the decision making process of the airline coordinators. In this paper we present the implementation of a Distributed Multi-AgentSystem (MAS) that represents the several roles that exist in an AOCC. This MAS deals with several operational bases and for each type of operation problems it has several specialized software agents that implements heuristic solutions and other solutions based in operations research mathematic models and artificial intelligence algorithms. These specialized agents compete to find the best solution for each problem. We present a real case study taken from an AOCC where a crew recovery problem is solved using the MAS. Computational results using a real airline schedule are presented, including a comparison with a solution for the same problem found by the human operators in the Airline Operations Control Center. We show that, even in simple problems and when comparing with solutions found by human operators in the case of this airline company, it is possible to find valid solutions, in less time and with a smaller cost.
Abstract. The Airline Operations Control Center (AOCC) tries to solve unexpected problems that might occur during the airline operation. Problems related to aircrafts, crewmembers and passengers are common and the actions towards the solution of these problems are usually known as operations recovery. In this paper we present the implementation of a Distributed Multi-AgentSystem (MAS) representing the existing roles in an AOCC. This MAS has several specialized software agents that implement different algorithms, competing to find the best solution for each problem and that include not only operational costs but, also, quality costs so that passenger satisfaction can be considered in the final decision. We present a real case study where a crew recovery problem is solved. We show that it is possible to find valid solutions, with better passenger satisfaction and, in certain conditions, without increasing significantly the operational costs.
Number of questions arises while building a Multiagentsystem such as kind of language used to communicate between agents, how conflicts are recognized and how to reach an agreement. How Autonomous agents achieve their goals. Multiagent systems are unique because agents can compute and process the information too. Each researcher has his own view about MULTIAGENTSYSTEM but the researchers have much in common. An example of MULTIAGENTSYSTEM is searching a answer for a specific question in internet is difficult like to get information about different flights. Instead of searching the details by visiting respective sites why can’t an agent do it on behalf of user? The agent is given a query to retrieve the desired information, the agent visits the different sites and displays the result to the use (Example is searching the flight details in makemytrip.com). Failure occurs if information is unavailable or because of network failure
Systems composed of interacting autonomous agents offer new ways in developing applica- tions in complex domains. Using a multi-agent platform to coordinate an information sys- tem is an appropriate choice because of the complexity and dynamism required. Data flux of an economical system is generally built to follow document movements. On the other hand, decision making and disseminating processes are complex and must be flexible and network distributed. Our goal is to build a Decision Support System (DSS) using JADE Multi-agentsystem. This paper reflects a small part of this goal so we are emphasizing the working plan coordination.
Agents are software programs that perform tasks on behalf of others and they can be used to mine data with their characteristics. Agents are task oriented with the ability to learn by themselves and they react to the situation. Learning characteristics of an agent is done by verifying its previous experience from its knowledgebase. An agent concept is a complementary approach to the Object Oriented paradigm with respect to the design and implementation of autonomous entities driven by beliefs, goals and plans. Email is one of the common means for communication via text. Email cleaning problem is formalized as non-text filtering and text normalization in a two step process. Agents incorporated in the architectural design of an Email data cleaning process combines both the features of Multi-AgentSystem (MAS) Framework and MAS with learning (MAS- L) Framework. MAS framework reduces the development time and the complexity of implementing the software agents. The MAS-L framework incorporates the intelligence and learning properties of software agents. MAS-L Framework makes use of the Decision Tree learning and an evaluation function to decide the next best decision that applies to the machine learning technique. This paper proposes the design for Multi- Agent based Data Cleaning Architecture that incorporates the structural design of agents into object model. The design of an architectural model for Multi-Agent based Data Cleaning inherits the features of the MAS and uses the MAS-L framework to design the intelligence and learning characteristics.
First experimental results suggest that the ABA’s learning capabilities were sufficient in order to beat its competitor (scenario 2). We observed on a 150 episodes experiment that the rough tendency was for the ABA to significantly increase the number of deals won over its competitor. However, the ABA was not successful in improving its utility on the deals it made. We believe this is due to the rewarding mechanism applied, which ranks the fact of getting a deal higher than the distinction between deal utilities. Furthermore, the exploration rate was not sufficient for the agent to try different tactic combinations in the episodes tested. The same difficulty appeared in scenario 1, where the ABA agent did not tend to increase the utility of its deals. This was also due to the fact that the use of conceder tactic combinations was preferred, since they lead faster to a deal and so their Q-values are increased in early stages of the adaptation process.
The communication between the distributed agents is asynchronous and done over Ethernet network using the TCP/IP protocol. The messages exchanged by the agents are encoded using the FIPA-ACL communication language, being their content formatted according to the FIPA-SL0 language. The meaning of the message content is standardized according to the designed GRACE ontology. The integration of the GRACE ontology, edited in Protégé, in the MAS solution was performed by using the OntologyBeanGenerator plug-in, which allows to automatically generate the Java classes from the Protégé tool, following the FIPA specifications. The main generated class represents the vocabulary and main ontological objects (i.e. concepts and predicates) defined in the ontology. The second group of generated Java classes specify the structure and semantics of each ontological object defined in the ontology. The integration of legacy systems, and particularly the interaction with physical devices hosted in the production line, assumes a critical role when deploying this kind of systems into industrial environments. As an example, QCA agents are associated to quality control functions, namely to the adaptation procedures allowing the improvement of the quality control station behaviour and consequently of the whole system behaviour. The interconnection between the QCA agent and the quality control station (QCS), illustrated in Fig. 8, comprises an intelligent part (the agent) and the physical part (the hardware device responsible for the inspection tasks), which in this work also embodies a measurement system developed in LabView™.
QoS have to be based on service performance, cost, availability, response time and also the trustworthiness of the agent. It will be possible to adapt the agent’s behaviors selection that will indirectly change the network topology to a more consistent one, structured with higher quality and better trustworthiness of the performed services. The assumptions on this approach allow joining several agents belonging to different societies after they make their mutual connections evolve. Due to the automatic creation of new relations, an agent can smoothly enter into a society. On the opposite side, the weak connections might be eliminated avoiding saturation of the system with useless agents. In this way, the system achieves equilibrium regarding the responsiveness from all societies, since the agents are able to evolve to accomplish the requests.
Demand for electricity should be made more adaptive to supply conditions, avoiding peaks of demand, resulting in a more efficient grid with lower prices for consumers. As a result, the new electrical grid intends to get an economic balance and increase the efficiency of the current the electri- cal supply. Energy efficient technologies such as intelligent controls systems that adjust the heating temperature, light- ing can help with the management of consumption in build- ings and houses. This intelligent control system can give consumers control over the amount of electricity they use. Furthermore, the intelligent control system can integrated into the power grid through equipment capable of collecting data about electricity consumption and of communicating with others entities in the power grid. A key element that allows all of the emerging smart grid technologies to func- tion together is the interactive relationship between the grid operators, utilities, and the user. Controls in the household and appliances can be set up to respond to signals from the energy grid to minimize the energy use at times when the power grid is under stress from high demand, or even to shift some of their power use to times when power is avail- able at a lower cost. This intelligent control system inside a household introduce the concept of Smart Home.
CANPLAN2  is a modular extension of CANPLAN , both BDI-based formal lan- guages that incorporate an HTN planning mechanism. One of the improvements in CANPLAN2 was to prevent an agent from blindly persisting with a blocked subgoal when an alternative plan is available for achieving a higher-level goal. This approach was further extended in  to ad- dress previous limitations such as failure handling, declarative goals, and lookahead planning. It is important to note that the CAN family are not implemented programming languages, although its features could be used to augment some BDI-based AOP languages. Similarly, in  the authors proposed an approach to obtain new abstract plans in BDI systems for hy- brid planning problems — where both goal states and the high-level plans already programmed are considered — bringing classical planning into BDI hierarchical structures. This approach was directed to single-agent planning in the context of AOP languages, it does not address the problems of multi-agent planning such as the ones that are approached in our work.
An approach has been proposed for the analysis, architectural design and formal verification of an Information Management System (IMS). A multi-agent based architecture is suitable for such a system. In a multi-agentsystem applications are designed in terms of autonomous software entities called agents that flexibly achieve their objectives by interacting with one another in terms of high level protocols and languages [Zambonelli, Jennings and Wooldridge, 2003]. An Agent is a self-contained program capable of controlling its own decision-making, based on its perception of environment, in pursuit of one or more objectives [Jennings and Wooldridge, 1996]. A method for the architecture and formal verification of university IMS has been proposed. The Belief, Desire, Intention (BDI) agent model [Bratman, 1987] has been adopted. Each agent is autonomous and can make decisions based on its knowledge-base. An agent based on BDI theory can adapt to changing situations by focusing on the most appropriate goal at the time [Rens, Ferrein and Van, 2009].
Note that in previous formula two propositions were used. Proposition HeatingIs Finished denotes that the system finishes the heating phase in the current step. Proposition CoolingIsStarted is true in the steps in which the system is in the cooling phase. The second model is MAS’ “structural” model, where the structure of multi-agentsystem is captured in particular ATS. Parts of MAS structure are agents, states of agents, transitions between agents’ states and propositions. Note that the set of propositions and the set of agents are the same in both models. ATL and ATS are approaches which assume weak definition of agent, and this fits quite well for this application. But, if AI have to be modeled, epistemic extension of these approaches will have to be used. Epistemic extensions are alternating- time epistemic temporal logic and alternating epistemic transition systems. Agents’ knowledge can be modeled in multiagent systems.
In fact, agent technology was begun in the 1950s. Agent is a software that user achieves automatically wanting work. In particular, this is a concept that has been studied for a long time in artificial intelligence. From the late 1980s, a boundary that is an agent has been detached with artificial intelligence and exposed to individual study subject. Agent products have appeared since the early 1990s  . A multi-agentsystem consists of multiple agents who are autonomous and make their decisions independently. By this definition, we rule out those systems where a central planner or designer controls the decision processes of local agents. If the agents’ actions do not affect each others’ outcomes, then we may as well consider the agents’ situations independently  . A multi-agentsystem offer certain advantages for problem solving: faster response,