Skip to main content
Top

2005 | Book

Design of Intelligent Multi-Agent Systems

Human-Centredness, Architectures, Learning and Adaptation

Authors: Nikhil Ichalkaranje, Professor Lakhmi C. Jain, Associate Professor Rajiv Khosla

Publisher: Springer Berlin Heidelberg

Book Series : Studies in Fuzziness and Soft Computing

insite
SEARCH

About this book

There is a tremendous interest in the design and applications of agents in virtually every area including avionics, business, internet, engineering, health sciences and management. There is no agreed one definition of an agent but we can define an agent as a computer program that autonomously or semi-autonomously acts on behalf of the user. In the last five years transition of intelligent systems research in general and agent based research in particular from a laboratory environment into the real world has resulted in the emergence of several phenomenon. These trends can be placed in three catego­ ries, namely, humanization, architectures and learning and adapta­ tion. These phenomena are distinct from the traditional logic­ centered approach associated with the agent paradigm. Humaniza­ tion of agents can be understood among other aspects, in terms of the semantics quality of design of agents. The need to humanize agents is to allow practitioners and users to make more effective use of this technology. It relates to the semantic quality of the agent design. Further, context-awareness is another aspect which has as­ sumed importance in the light of ubiquitous computing and ambi­ ent intelligence. The widespread and varied use of agents on the other hand has cre­ ated a need for agent-based software development frameworks and design patterns as well architectures for situated interaction, nego­ tiation, e-commerce, e-business and informational retrieval. Fi- vi Preface nally, traditional agent designs did not incorporate human-like abilities of learning and adaptation.

Table of Contents

Frontmatter
Chapter 1. Humanization of Soft Computing Agents
Abstract
Soft computing agents today are being applied in a range of areas including image processing, engineering, process control, data mining, internet and others. In the process of applying soft computing agents to complex real world problems three phenomena have emerged. Firstly, the application of soft computing agents in distributed environments has resulted in merger of techniques from soft computing area with those in distributed artificial intelligence. Secondly, in order to facilitate better understanding and frequent use of the soft computing agent technology, humanization of these agents has become an important design issue (Takagi, 2001). Thirdly, given the approximate and imprecise nature of the solutions provided by soft computing agents, optimization has become an important design issue in an effort to improve the quality of solution provided by soft computing agents. Finally, from a user’s perspective, as the soft computing agent technologies have moved out of laboratories into the internet and the real world the need to model and express these technologies in the problem solving context of user. In this chapter we model these four phenomenons as part of multi-layered multi-agent architecture. The layered architec­ture is also motivated by the human-centered approach and criteria outlined in the 1997 NSF workshop on human-centered systems.
Rajiv Khosla, Qiubang Li, Chris Lai
Chapter 2. Software Agents for Ubiquitous Computing
Abstract
Ubiquitous computing draws a picture of a world where people access intelligent and easy-to-use services, irrespective of location and time. Software agent technology is a good candidate for ubiquitous computing, because the agent paradigm is by definition well suited for decentralized, heterogeneous and dynamic environments. This chapter presents the general requirements of the ubiquitous environment, and discusses the use of software agent technology to facilitate service provision and composition for ubiquitous client end systems.
Sasu Tarkoma, Mikko Laukkanen, Kimmo Raatikainen
Chapter 3. Agents-Based Knowledge Logistics
Abstract
A research carried out in the framework of the knowledge logistics lies in the base of the chapter. As a result of the research an approach addressing the knowledge logistics problem was developed. The approach considers the problem as a problem of configuring a knowledge source network that is assumed to consist of distributed heterogeneous knowledge sources. An implementation of the approach was put into practice through its realization in the system “KSNet”. Distribution and heterogeneity of the knowledge sources determine a distributed and scalable character to the problem of the network configuring. Such nature of the problem causes for the system to have a multi-agent architecture. This chapter presents a prototype of the developed agent community implementation in the system “KSNet” and a constraint-based protocol designed for the agents’ negotiation. An application of the developed agent community to coalition-based operations support and the protocol are illustrated via case studies of a mobile hospital configuration as a task of health service logistics and automotive supply network configuration.
Alexander Smirnov, Mikhail Pashkin, Nikolai Chilov, Tatiana Levashova
Chapter 4. Architectural Styles and Patterns for Multi-Agent Systems
Abstract
A Multi-Agent System (MAS) is an organization of coordinated autonomous agents that interact in order to achieve common goals. Considering real world social organizations as an analogy (Zambonelli et al. 2000), this chapter proposes architectural styles and design patterns for MAS which adopt concepts from social theories. The styles are intended to represent a macro-level architecture of a MAS in terms of actor, goal and actor dependency and are evaluated with respect to software quality attributes. At a micro-level, social patterns give a finer-grain description of the MAS architecture and define how goals assigned to agents will be fulfilled. They are modeled within a conceptual framework analyzing them from five points of view: social, intentional, structural, communicational and dynamic. An e-business example illustrates our purpose.
Manuel Kolp, T. Tung Do, Stéphane Faulkner, T. T. Hang Hoang
Chapter 5. Design and Behavior of a Massive Organization of Agents
Abstract
Researches in the domain of multi-agent systems have currently a very important development. The plastic, dynamic and distributed characters of the these systems permit several advantages: in simulation, they can easily express the behavior of real systems for which equational models are insufficient and, in effective working, they allow to produce some typically adaptive behaviors. Multiagent systems are used in varied domains where they present an interesting alternative to the classical approaches. But there exists a limitation to their utilization: one doesn’t know how easily conceive or cleverly control the behavior of multi-agent systems of very large size, containing about ten or hundred thousand agent (Huhns & Singh 1998).
Alain Cardon
Chapter 6. Developing Agent-Based Applications with JADE
Abstract
JADE (Java Agent Development Framework) is an “open source” FIPA-compliant software environment to build agent systems. JADE offers an agent middleware to implement efficient FIPA2000 compliant multi-agent systems and supports their development through the availability of a predefined programmable agent model, an ontology development support, and a set of management and testing tools. This chapter describes JADE and its use in three international projects to develop applications in the fields of: corporate memory management, integration of fixed and mobile networks, and integration of Web services.
F. Bergenti, A. Poggi, G. Rimassa, P. Turci, M. Tomaiuolo
Chapter 7. A Collective Can Do Better
Abstract
Can we devise simple solutions to complex problems? Is it possible to do so by making use of elemental modules, which when collaborating create emerging intelligence? The answer is yes. No complex mathematical models are required. Nature offers a variety of techniques that lend themselves well to solving complex problems making use of simpler atomic entities. Insects, for instance, as individuals are very simple, but in a collective they are powerful systems able to solve very complex tasks. This chapter describes how the insect world can inspire engineers and computer scientists to devise simple solutions to complex problems. After all the simplest solution is always the best.1
N. D. Monekosso, P. Remagnino
Chapter 8. Coordinating Multi-Agent Assistants with an Application by Means of Computational Reflection
Abstract
Assistant agents are employed to provide users working with an application with various help services aimed at, e.g., organising user data, performing useful actions on the application on users’ behalf, and suggesting suitable, independently obtained, information. In order to better structure the implementation of the (frequently heterogeneous) assistance functionalities, it would often be desirable to introduce, in lieu of a monolithic agent, a group of independent ones, each specialised for a simple, well-defined task. Such assistant agents need to interact with each other and with the application they extend, for the sake of exchanging data and coordinating their tasks.
A. Di Stefano, G. Pappalardo, C. Santoro, E. Tramontana
Chapter 9. Learning by Exchanging Advice
Abstract
The emergence of Multiagent systems brought new challenges to the field of Machine Learning, as it did to many others. One of the main challenges is to take advantage of the information available when several agents, possibly using different learning techniques, are dealing with similar problems, either in the same location (i.e. acting as a team) or in different ones. This work aims at studying the possible advantages and pitfalls of exchanging information during the learning process, leading to better adaptation. We will discuss the subject of when, how and to whom ask for advice, and present the results obtained in two experimental scenarios: the Pursuit (Predator-Prey) Domain and a Traffic Control simulation. Results show that exchange of information can improve the average performance of learning agents enabling them to escape from local maxima in some cases, although it may reduce the exploration of the space, preventing successful agents from finding better local maxima of the quality function.
Eugénio Oliveira, Luís Nunes
Chapter 10. Adaptation and Mutation in Multi-Agent Systems and Beyond
Abstract
Reconfigurable and mutable systems are increasingly more popular. As early as 1975, the Microsoft Basic interpreter for Altair contained self-modifying code, introduced to overcome resource limitations (only 4K of space available for the interpreter). A contemporary web browser is a custom application, consisting of a basic framework with multiple extension API’s and a large number of plug-ins, codecs, drivers, applets, controls, themes and other addons. These extensions are usually developed by third parties, installed/uninstalled dynamically during the lifetime of the application, and frequently changing the behavior of the application in a radical way. Some of the changes in functionality are desired, or at least approved by the user: an example of such an extension is the ability to view new media formats. Frequently, some of the effects are undesirable from the user’s point of view: some third party extensions contain spyware,pieces of code which report usage statistics and other information about the user. Occasionally, viruses and worms use the very same extension API’s.
Ladislau Bölöni, Dan Cristian Marinescu
Chapter 11. Intelligent Action Acquisition for Animated Learning Agents
Abstract
Generation of animated human figures especially in crowd scenes has many applications in such domains as the special effects industry, computer games or for the simulation of the evacuation from crowded areas. Current systems allow for partially automatic generation of scenes involving a few interacting characters but expensive manual labour is still necessary in order to enrich the characters’ behaviour repertoire. In this chapter we explore the possibility of applying reinforcement learning to acquire new high-level actions for animated characters. The chosen algorithm is the deterministic version of Q-learning. This allows for easy definition of the task, since only the ultimate goal of the learning agent must be defined. Generated actions can then be used to enrich the animation produced by an animation system. Results achieved when training agents with forward and inverse kinematics control are also demonstrated and compared.
Adam Szarowicz, Marek Mittmann, Jaroslaw Francik
Chapter 12. Using Stationary and Mobile Agents for Information Retrieval and E-Commerce
Abstract
The deployment and widespread use of Internet generated a renewed interest in distributed architectures. However, a great number of services offered with these architectures require high-speed network connections. Moreover, there is a widespread proliferation of portable computers and devices (e.g., laptops, palmtops, etc.) which are equipped with low processing capacity. Thus, there is a need for a new service engineering that is able to deal with both the demands of high-speed network connections and the capacity limits of the new portable information devices. This new service engineering could very well benefit from multi-agent systems and mobile agents, especially when it comes to the realms of information retrieval and e-commerce. This chapter presents a synthesis of agent technologies and some of its novel applications. Section 1 introduces the basic concepts related to agent technologies. Section 2 presents an agent-based architecture for information retrieval. Section 3 provides the implementation details of this architecture, whose performance is assessed in Section 4. Section 5 suggests a new architecture for product search in e-commerce. Also, Section 6 concludes by highlighting the salient features of agent-based architectures.
Samuel Pierre
Metadata
Title
Design of Intelligent Multi-Agent Systems
Authors
Nikhil Ichalkaranje
Professor Lakhmi C. Jain
Associate Professor Rajiv Khosla
Copyright Year
2005
Publisher
Springer Berlin Heidelberg
Electronic ISBN
978-3-540-44516-6
Print ISBN
978-3-642-06177-6
DOI
https://doi.org/10.1007/978-3-540-44516-6

Premium Partner