Dynamic Filtering of Useless Data in an Adaptive Multi-Agent System: Evaluation in the Ambient Domain

Guivarch, Valérian; Camps, Valérie; Péninou, André; Stuker, Simon

doi:10.1007/978-3-642-38073-0_10

Cited by 4 publications

(3 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For instance, the extension of the MAS Amadeus in [15] provides capabilities to learn the user's behavior. Some more general works try to provide formal descriptions of knowledge for some tasks and procedures to apply it.…”

Section: Management Of Context-aware Workflowsmentioning

confidence: 99%

Implementation of context-aware workﬂows with multi-agent systems

et al. 2016

View full text Add to dashboard Cite

Cite this article as: Javier Alfonso-Cendón, José M. Fernández-de-Alba, Rubén Fuentes-Fernández, Juan Pavón, Implementation of context-aware workflows with Multi-agent Systems, Neurocomputing, http://dx.doi.org/10.1016/j. neucom.2014.10.098 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Abstract. Systems in Ambient Intelligence (AmI) need to manage workflows that represent users' activities. These workflows can be quite complex, as they may involve multiple participants, both physical and computational, playing different roles. Their execution implies monitoring the development of the activities in the environment, and taking the necessary actions for them and the workflow to reach a certain end. The context-aware approach supports the development of these applications to cope with event processing and regarding information issues. Modeling the actors in these context-aware workflows, where complex decisions and interactions must be considered, can be achieved with multi-agent systems. Agents are autonomous entities with sophisticated and flexible behaviors, which are able to adapt to complex and evolving environments, and to collaborate to reach common goals. This work presents architectural patterns to integrate agents on top of an existing context-aware architecture. This allows an additional abstraction layer on top of context-aware systems, where knowledge management is performed by agents. This approach improves the flexibility of AmI systems and facilitates their design. A case study on guiding users in buildings to their meetings illustrates this approach.

show abstract

Section: Management Of Context-aware Workflowsmentioning

confidence: 99%

Implementation of context-aware workﬂows with multi-agent systems

et al. 2016

View full text Add to dashboard Cite

show abstract

“…Sa résolution s'appuie sur un processus local impliquant l'agent donnée et les différents agents contexte de son voisinage, afin d'évaluer non pas l'inutilité de la donnée, mais plutôt son utilité. L'objectif recherché est alors d'estimer le niveau d'utilité de la donnée en fonction des différents effecteurs, pour ensuite considérer comme inutile une donnée présentant un niveau d'utilité trop faible vis-à-vis d'un effecteur, et ainsi prévenir l'apparition de cette SNC (Guivarch et al, 2013).…”

Section: Agent Donnéeunclassified

Apprentissage comportemental en continu de dispositifs ambiants par systèmes multi-agents adaptatifs

Guivarch¹,

Camps²,

Péninou³

et al. 2015

Revue d'intelligence artificielle

Self Cite

View full text Add to dashboard Cite

Nous proposons d'étendre la résolution émergente de problème au cadre des systèmes ambiants, au travers de la conception du système multi-agent adaptatif Amadeus. L'objectif d'Amadeus est d'apprendre le comportement correct à attribuer à un système ambiant. Cet apprentissage passe par l'observation des différentes actions des utilisateurs, afin d'être capable de progressivement réaliser ces actions à leur place. Pour être réellement adaptatif, Amadeus réalise un apprentissage en continu sans connaissance a priori ; il s'adapte à l'apparition/disparition de dispositifs et est capable de modifier son comportement en cours de fonctionnement en cas d'évolution du comportement utilisateur. Amadeus a été évalué par simulation sur plusieurs scénarios, et a montré sa capacité à apprendre localement le comportement correct à attribuer à un dispositif par l'observation des actions utilisateur.ABSTRACT. We propose to extend the emergent problem solving to the ambient system field, through the design of the adaptive multi-agent system Amadeus. The goal of Amadeus is to learn the correct behavior to assign to an ambient system. This learning is based on the observation of users actions, in order to be able to progressively perform these actions on their behalf. To be truly adaptive, Amadeus performs a continuous learning without any a priori knowledge; it adapts itself to the appearance/disappearance of devices and it is able to modify its behaviour at runtime if there is a change in the users behavior. Amadeus has been evaluated by simulation through several scenarios, and has demonstrated its capability to learn in a local way the correct behavior to assign to a device by observing users actions.MOTS-CLÉS : systèmes multi-agents adaptatifs, apprentissage, systèmes ambiants.

show abstract

“…management of smart cities and the appearance of new transport modes and behavior). Dynamic problems are often associated with ambient environments in which new devices may appear or disappear dynamically [12,16]. Many classifiers are themselves composed of multiple linear classifiers (i.e.…”

Section: Introductionmentioning

confidence: 99%

How to Solve a Classification Problem Using a Cooperative Tiling Multi-agent System?

Thibault

Verstaevel

Migeon

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Adaptive Multi-Agent Systems (AMAS) transform dynamic problems into problems of local cooperation between agents. We present smapy, an ensemble based AMAS implementation for mobility prediction, whose agents are provided with machine learning models in addition to their cooperation rules. With a detailed methodology, we propose a framework to transform a classification problem into a cooperative tiling of the input variable space. We show that it is possible to use linear classifiers for online non-linear classification on three benchmark toy problems chosen for their different levels of linear separability, if they are integrated in a cooperative Multi-Agent structure. The results obtained show a significant improvement of the performance of linear classifiers in non-linear contexts in terms of classification accuracy and decision boundaries, thanks to the cooperative approach.

show abstract

Dynamic Filtering of Useless Data in an Adaptive Multi-Agent System: Evaluation in the Ambient Domain

Cited by 4 publications

References 9 publications

Implementation of context-aware workﬂows with multi-agent systems

Implementation of context-aware workﬂows with multi-agent systems

Apprentissage comportemental en continu de dispositifs ambiants par systèmes multi-agents adaptatifs

How to Solve a Classification Problem Using a Cooperative Tiling Multi-agent System?

Contact Info

Product

Resources

About