Ramon Sangüesa

Extracting reputation in multi agent systems by means of social network topology

The problem of calculating a degree of reputation for agents acting as assistants to the members of an electronic community is discussed and a solution presented. Usual reputation mechanisms rely on feedback after interaction between agents. An alternative way to establish reputation is related with the position of each member of a community within the corresponding social network. We propose a method based on this idea, which is also used by well-known ranking algorithms, discuss its properties as well as experimental results and compare them to other reputation mechanisms for electronic communities supported by agents. The method proposed uses only local information in order to extract reputation and it is able to adapt automatically to the topology of the network or graph.

Possibilistic conditional independence: A similarity-based measure and its application to causal network learning

Abstract A definition for similarity between possibility distributions is introduced and discussed as a basis for detecting dependence between variables by measuring the similarity degree of their respective distributions. This definition is used to detect conditional independence relations in possibility distributions derived from data. This is the basis for a new hybrid algorithm for recovering possibilistic causal networks. The algorithm POSSCAUSE is presented and its applications discussed and compared with analogous developments in possibilistic and probabilistic causal networks learning.

Tag Mechanisms Evaluated for Coordination in Open Multi-Agent Systems

Tags are arbitrary social labels carried by agents. When agents interact preferentially with those sharing the same Tag, groups are formed around similar Tags. This property can be used to achieve desired group coordination by evolving agents Tags through a group selection process. In this paper Tags performance is for the first time compared by simulation with alternative mechanisms for coordinated learning in multi-agent systems populations. We target open systems, hence we do not make costly assumptions on agent capabilities (rational or computational). It is a requirement that coordination strategies prove simple to implement and scalable. We build a simulator incorporating competition and cooperation scenarios modeled as one-shot repeated games between agents. Tags prove to be a very good coordination mechanism in both, cooperation building in competitive scenarios and agent behavior coordination in fully cooperative scenarios.

Application of Bayesian Network Learning Methods to Waste Water Treatment Plants

Bayesian Networks have been proposed as an alternative to rule-based systems in domains with uncertainty. Applications in monitoring and control can benefit from this form of knowledge representation. Following the work of Chong and Walley, we explore the possibilities of Bayesian Networks in the Waste Water Treatment Plants (WWTP) monitoring and control domain. We show the advantages of modelling knowledge in such a domain by means of Bayesian networks, put forth new methods for knowledge acquisition, describe their applications to a real waste water treatment plant and comment on the results. We also show how a Bayesian Network learning environment was used in the process and which characteristics of data in the domain suggested new ways of representing knowledge in network form but with uncertainty representations formalisms other than probability. The results of applying a possibilistic extension of current learning methods are also shown and compared.

A parallel algorithm for building possibilistic causal networks

Abstract Among the several representations of uncertainty, possibility theory allows also for the management of imprecision coming from data. Domain models with inherent uncertainty and imprecision can be represented by means of possibilistic causal networks that, the possibilistic counterpart of Bayesian belief networks. Only recently the definition of possibilistic network has been clearly stated and the corresponding inference algorithms developed. However, and in contrast to the corresponding developments in Bayesian networks, learning methods for possibilistic networks are still few. We present here a new approach that hybridizes two of the most used approaches in uncertain network learning: those methods based on conditional dependency information and those based on information quality measures. The resulting algorithm, POSSCAUSE, admits easily a parallel formulation. In the present paper POSSCAUSE is presented and its main features discussed together with the underlying new concepts used.

A Ranking Algorithm Based on Graph Topology to Generate Reputation or Relevance

We introduce NodeRanking, a new mechanism for ranking the importance of nodes in a graph. Ranking nodes in a graph is not a new problem and some solutions do exist in the literature, such as the well-known algorithms Pagerank, HITS, and SALSA. Our main contribution consists in the fact that the introduced algorithm uses only local information. Thus, the algorithm is truly distributed and it does not need any knowledge of the whole graph. Furthermore, NodeRanking adapts itself to the graph topology, hence, no setting-up process is required. The quality and adaptability of NodeRanking is tested on different graphs (a real social network and a scale-free graph) with different topological properties Finally, we show how the algorithm may be applied either to extract the relevance of Web pages or to infer the reputation of members of a community.

Design, Implementation, and Evaluation of a Resource Management Multiagent System for a Multimedia Processing Grid

This paper proposes marketplace-based agent architecture for Grid Resource Management that works on the application level. Relaying in FIPA as agent systems standard and Globus Toolkit as “de facto” middleware solution for Grid computing, we achieve the desired flexibility and modularity in order to provide a pluggable agent layer for a broader generic Grid Computing framework. The agents in the system use a utility table built previously to the system operation as information source for improving their negotiation abilities. It allows them to, given a state of the resources on the Grid, check the predicted performance of the possible configurations and select the best-rated values for some task execution parameters. We have implemented the architecture using JADE and have performed preliminary experiments consisting on multimedia processing for the conversion between video formats.

Video transcoding in a grid network with user controlled lightpaths

GridON is an application that converts high-resolution broadcast video into MPEG-2 format, hereby reducing the file size and resolution. The application uses the user controlled lightpaths (UCLP) software to create on-demand, end-to-end, high-bandwidth dedicated connections to access remote computers. The converted MPEG-2 files can be distributed much faster and further than the source files to these dispersed computers, for reassembly into the higher resolution format. This paper describes the demonstration that took place last September at the iGrid 2005 conference held in San Diego. As a proof of concept, we successfully demonstrated that a video transfer in a Grid network environment can be integrated with a user-controlled lightpath provisioning system.

A Group Selection Pattern Applied to Grid Resource Management

A key challenge in grid computing is the achievement of efficient and self-organized resource management. Grids are often large scale, heterogeneous, and unpredictable systems. Introducing group structures can help to distribute coordination efforts, but higher levels of adaptation and learning in the coordination protocols are still required in order to cope with system complexity. We provide a solution based on a self-organized and emergent mechanism evolving congregations of resource management agents through a group selection process which maximizes utility outcomes for system-wide performance. We provide a formalization of this process into a group selection pattern, and we propose several instantiations optimizing grid resource management scenarios such as adaptive job scheduling, market-based resource management, and policy coordination in virtual organizations (VOs). We further evaluate by simulation the performance of the mechanism in those scenarios. The results support the conclusion that group selection optimizes coordination by evolving small and dynamic groups.

A group selection pattern optimizing job scheduling in decentralized grid markets

Decentralized economic models are being considered as scalable coordination mechanism for the management of service allocations to clients. However, decentralization incorporates further dynamicity and unpredictability into the system, degrading its performance. In this paper, a solution based on a self-organized and emergent Group Selection mechanism is proposed. Dynamic congregations evolve Grid Markets participants (c and service providers) into optimized market segments, maximizing utility outcomes for system-wide performance. We provide evaluation by simulation of the Group Selection mechanism performance in a market-based resource management and job scheduling scenario for Grid computing, compared with alternative scheduling strategies such as economic in a flat population (not using groups), random and least loaded resource selection.