基于多智能体强化学习的乳腺癌致病基因预测

通过分析基因突变过程, 提出利用强化学习对癌症患者由正常状态至患病状态的过程进行推断, 发现导致患者死亡的关键基因突变. 首先, 将基因视为智能体, 基于乳腺癌突变数据设计多智能体强化学习环境; 其次, 为保证智能体探索到与专家策略相同的策略和满足更多智能体快速学习, 根据演示学习理论, 分别提出两种多智能体深度Q网络: 基于行为克隆的多智能体深度Q网络和基于预训练记忆的多智能体深度Q网络; 最后, 根据训练得到的多智能体深度Q网络进行基因排序, 实现致病基因预测. 实验结果表明, 提出的多智能体强化学习方法能够挖掘出与乳腺癌发生、发展过程密切相关的致病基因.     

Reflections on the Nature of Multi-Agent Coordination and Its Implications for an Agent Architecture

The development of enabling infrastructure for the next generation of multi-agent systems consisting of large numbers of agents and operating in open environments is one of the key challenges for the multi-agent community.Current infrastructure support does not materially assist in the development of sophisticated agent coordination strategies. It is the need for and the development of such a high-level support structure that will be the focus of this paper. A domain-independent (generic) agent architecture is proposed that wraps around an agent's problem-solving component in order to make problem solving responsive to real-time constraints, available network resources, and the need to coordinate—both in the large and small—with problem-solving activities of other agents. This architecture contains five components, local agent scheduling, multi-agent coordination, organizational design, detection and diagnosis, and on-line learning, that are designed to interact so that a range of different situation-specific coordination strategies can be implemented and adapted as the situation evolves. The presentation of this architecture is followed by a more detailed discussion on the interaction among these components and the research questions that need to be answered to understand the appropriateness of this architecture for the next generation of multi-agent systems.     

XML在多代理系统中的应用

随着软件系统复杂度的不断提高,在软件开发应用中多代理系统越来越显示出重要的作用。多代理系统的通信语言的研究也就越来越迫切。该文讨论了多代理系统的通信语言,提出以XML作为多代理系统的通信语言,并对此做了探讨。     

Compositional Verification of a Multi-Agent System for One-to-Many Negotiation

Verification of multi-agent systems hardly occurs in design practice. One of the difficulties is that required properties for a multi-agent system usually refer to multi-agent behaviour which has nontrivial dynamics. To constrain these multi-agent behavioural dynamics, often a form of organisational structure is used, for example, for negotiating agents, by following strict protocols. The claim is that these negotiation protocols entail a structured process that is manageable with respect to analysis, design and execution of such a multi-agent system. In this paper this is shown by a case study: verification of a multi-agent system for one-to-many negotiation in the domain of load balancing of electricity use. A compositional verification method for multi-agent systems is applied that allows to (1) logically relate dynamic properties of the multi-agent system as a whole to dynamic properties of agents, and (2) logically relate dynamic properties of agents to properties of their subcomponents. Given that properties of these subcomponents can be verified by more standard methods, these logical relationships provide proofs of the dynamic properties of the multi-agent system as a whole.     

MOBMAS: A methodology for ontology-based multi-agent systems development

Ontologies offer significant benefits to multi-agent systems: interoperability, reusability, support for multi-agent system (MAS) development activities (such as system analysis and agent knowledge modeling) and support for MAS operation (such as agent communication and reasoning). This paper presents an ontology-based methodology, MOBMAS, for the analysis and design of multi-agent systems. MOBMAS is the first methodology that explicitly identifies and implements the various ways in which ontologies can be used in the MAS development process and integrated into the MAS model definitions. In this paper, we present comprehensive documentation and validation of MOBMAS.     

A security-aware metamodel for multi-agent systems (MAS)

This paper adopts a model-based security (MBS) approach to identify security requirements during the early stages of multi-agent system development. Our adopted MBS approach is underpinned by a metamodel independent of any specific methodology. It allows for security considerations to be embedded within any situated agent methodology which then prescribes security considerations within its work products. Using a standard model-driven engineering (MDE) approach, these work products are initially constructed as high abstraction models and then transformed into more precise models until code-specific models can be produced. A multi-agent system case study is used to illustrate the applicability of the proposed security-aware metamodel.     

Temporal Development Methods for Agent-Based

In this paper we overview one specific approach to the formal development of multi-agent systems. This approach is based on the use of temporal logics to represent both the behaviour of individual agents, and the macro-level behaviour of multi-agent systems. We describe how formal specification, verification and refinement can all be developed using this temporal basis, and how implementation can be achieved by directly executing these formal representations. We also show how the basic framework can be extended in various ways to handle the representation and implementation of agents capable of more complex deliberation and reasoning.This revised version was published online in August 2005 with a corrected cover date.     

A formal framework for multi-agent systems analysis and design

Agent based computing is generally intended for modeling and implementation of distributed complex problems. Despite the existence of many applications, the problem of rational engineering of multi-agent systems remains complex and difficult. The purpose of this paper can be summarized within two claims. First, we aim at providing an approach that gives some guidelines for specifying and designing multi-agent systems. Secondly, we focus on the formalisms as a language for describing the models produced in each development process phases. These seem to be straightforward, while the development of multi-agent systems is still done, in most cases, without using methods and formal modeling such as those generally used in object oriented software. We illustrate this approach by specifying an example based upon a specific agent architecture.     

多主体行为模拟的层次Petri网方法

对多主体系统中主体行为间关系进行准确的描述和分析是保证多主体系统顺利运行的关键因素之一。与已有的基于Petri网的方法不同,该文将主体的动作分类为内部和外部动作,利用Petri网系统描述单个主体的内部结构,进而得到多主体系统的层次Petri网系统模型,每个主体的内部结构在系统网系统中表现为一个替代变迁元素,如此可以从两个层次（系统层和主体层）上对多主体系统的各种动态行为间的关系,特别是并发关系进行描述和分析。     

Organizational structures supported by agent-oriented methodologies

Agent technology is a software paradigm that permits to implement large and complex distributed applications. In order to assist the development of multi-agent systems, agent-oriented methodologies (AOM) have been created in the last years to support modelling more and more complex applications. Even though agents are perceived as autonomous entities that act according to some objectives, they are also members of a society, and have to exchange information with other agents and maintain some relationships at an organizational level. Modern AOMs should be able to capture and represent organizational structures, defining interaction and collaboration patterns between agents, their internal roles and dependencies between groups of agents. This paper analyses the most notable AOMs, paying attention to the support and possibilities that they offer for modelling organizational structures with different levels of complexity. This work can help developers to select the most appropriate methodology taking into account the social and organizational requirements of the multi-agent system to be deployed.     

A two-layered multi-agent reinforcement learning model and algorithm

Multi-agent reinforcement learning technologies are mainly investigated from two perspectives of the concurrence and the game theory. The former chiefly applies to cooperative multi-agent systems, while the latter usually applies to coordinated multi-agent systems. However, there exist such problems as the credit assignment and the multiple Nash equilibriums for agents with them. In this paper, we propose a new multi-agent reinforcement learning model and algorithm LMRL from a layer perspective. LMRL model is composed of an off-line training layer that employs a single agent reinforcement learning technology to acquire stationary strategy knowledge and an online interaction layer that employs a multi-agent reinforcement learning technology and the strategy knowledge that can be revised dynamically to interact with the environment. An agent with LMRL can improve its generalization capability, adaptability and coordination ability. Experiments show that the performance of LMRL can be better than those of a single agent reinforcement learning and Nash-Q.     

The multi-agent programming contest from 2005–2010

The Multi-Agent Programming Contest is an annual international event on programming multi-agent systems: Teams of agents participate in a simulated cooperative scenario. It started in 2005 and is organised in 2010 for the sixth time. The contest is an attempt to stimulate research in the area of multi-agent system development and programming by (i) identifying key problems in the field and (ii) collecting suitable benchmarks that can serve as milestones for testing multi-agent programming languages, platforms and tools. This article provides a short history of the contest since it started and reports in more detail on the cows and cowboys scenario implemented for the 2008, 2009 and 2010 contest editions. We briefly discuss the underlying technological background and conclude with a critical discussion of the experiences and lessons learned.     

Connectivity-maintaining Consensus of Multi-agent Systems With Communication Management Based on Predictive Control Strategy

This paper studies the connectivity-maintaining consensus of multi-agent systems.Considering the impact of the sensing ranges of agents for connectivity and communication energy consumption,a novel communication management strategy is proposed for multi-agent systems so that the connectivity of the system can be maintained and the communication energy can be saved.In this paper,communication management means a strategy about how the sensing ranges of agents are adjusted in the process of reachin...     

Improving the security level of the FUSION@ multi-agent architecture

The use of architectures based on services and multi-agent systems has become an increasingly important part of the solution set used for the development of distributed systems. Nevertheless, these models pose a variety of problems with regards to security. This article presents the Adaptive Intrusion Detection Multi-agent System (AIDeMaS), a mechanism that has been designed to detect and block malicious SOAP messages within distributed systems built by service based architectures. AIDeMaS has been implemented as part of FUSION@, a multi-agent architecture that facilitates the integration of distributed services and applications to optimize the construction of highly-dynamic multi-agent systems. One of the main features of AIDeMaS is that is employs case-based reasoning mechanisms, which provide it with great learning and adaptation capabilities that can be used for classifying SOAP messages. This research presents a case study that uses the ALZ-MAS system, a multi-agent system built around FUSION@, in order to confirm the effectiveness of AIDeMaS. The preliminary results are presented in this paper.     

Dynamic explicitly specified behaviors in distributed agent-based industrial solutions

Currently, the manufacturing domain is primarily characterized by the flexibility, adaptability and robustness of the production system. The manufacturing flow processes lead to shorter cycle times to efficiently meet customer needs. Mentioned features can be more easily achieved in a distributed system, such as holonic or multi-agent system, which becomes strongly influenced by the advancement of semantic technologies. In the majority of existing multi-agent based control systems, which are responsible for acting, sensing, computing and production planning, the ontology (necessary for knowledge bases and communication) is usually hard-coded directly in the agent code. In this case, the hard-coded system behavior can be hardly maintained??usually system reprogramming is needed from time to time to satisfy customer requirements. In this paper we discuss the necessity of explicit definition of both declarative and procedural knowledge and propose explicit procedural knowledge handling. Sharing and distribution of such knowledge is discussed and illustrated on an implemented transportation system example. We also introduce the utilization of discussed architecture for explicit specification of agent behavior in failures patterns handling and smart grid configuration scenario. Such a solution greatly increases the possibility of system integration, openness, flexibility, and extensibility, all without having to restart the running distributed system. The topic discussed in this paper shows the ability of the dynamic reconfigurable multi-agent system to participate in development of industrial control systems and solutions.     

基于值分布的多智能体分布式深度强化学习算法

近年来深度强化学习在一系列顺序决策问题中取得了巨大的成功,使其为复杂高维的多智能体系统提供有效优化的决策策略成为可能.然而在复杂的多智能体场景中,现有的多智能体深度强化学习算法不仅收敛速度慢,而且算法的稳定性无法保证.本文提出了基于值分布的多智能体分布式深度确定性策略梯度算法(multi-agent distribut...     

An autonomous agent development environment for engineering applications

Agent technology is playing an increasingly important role in developing distributed, intelligent, and collaborative engineering applications. The innate difficulties of interoperation between heterogeneous agent communities and rapid development of multi-agent systems have motivated the emergence of FIPA specifications and the proliferation of multi-agent system platforms or toolkits that implement FIPA specifications. This paper presents a FIPA compliant multi-agent framework called Autonomous Agent Development Environment (AADE). Originating from the engineering fields, AADE can facilitate the rapid development of collaborative engineering applications (especially in engineering design and manufacturing fields) through the provision of reusable packages of agent level components and programming tools. An agent oriented engineering project on the development of a collaborative product design environment is implemented based on the facilities provided by AADE.     

多智能体多目标跟踪算法研究

研究了多智能体同时跟踪多个目标的群集算法,采用智能体的编号对目标数求余数的方法来确定各个智能体的跟踪目标。对于势场函数,根据智能体目标是否相同建立了两种势场函数,并在此基础上确定动态系统的控制输入,使相同目标的智能体在跟踪目标的过程中形成群集,而不同目标的智能体相互分离;用李亚普诺夫稳定性理论分析了算法的收敛性。仿真结果表明,所提算法可有效解决多目标跟踪问题。     

Distributed Computation Multi-agent System

This article addresses a formal model of a distributed computation multi-agent system. This model has evolved from the experimental research on using multi-agent systems as a ground for developing fuzzy cognitive maps. The main paper contribution is a distributed computation multi-agent system definition and mathematical formalization based on automata theory. This mathematical formalization is tested by developing distributed computation multi-agent systems for fuzzy cognitive maps and artificial neural networks – two typical distributed computation systems. Fuzzy cognitive maps are distributed computation systems used for qualitative modeling and behavior simulation, while artificial neural networks are used for modeling and simulating complex systems by creating a non-linear statistical data model. An artificial neural network encapsulates in its structure data patterns that are hidden in the data used to create the network. Both of these systems are well suited for formal model testing. We have used evolutionary incremental development as an agent design method which has shown to be a good approach to develop multi-agent systems according to the formal model of a distributed computation multi-agent system.     

A taxonomy for multi-agent robotics

A key difficulty in the design of multi-agent robotic systems is the size and complexity of the space of possible designs. In order to make principled design decisions, an understanding of the many possible system configurations is essential. To this end, we present a taxonomy that classifies multi-agent systems according to communication, computational and other capabilities. We survey existing efforts involving multi-agent systems according to their positions in the taxonomy. We also present additional results concerning multi-agent systems, with the dual purposes of illustrating the usefulness of the taxonomy in simplifying discourse about robot collective properties, and also demonstrating that a collective can be demonstrably more powerful than a single unit of the collective.The support of NSERC Canada is greatfully acknowledged.