|
Publication accepted EUMAS2011 |
 |
 |
 |
|
Recently we have reported acceptance of an article for oral presentation at the next edition of the congress EUMAS for the 2011 edition: http://swarmlab.unimaas.nl/eumas2011/ The article title is "A Modular Architecture for Optimising Agent Hypothesis confirmation cost in Network Diagnosis" Alvaro Carlos Carrera and A. Iglesias. The article proposes a flexible architecture that allows adding or subtracting some modules to increase the autonomy or reduce the computational cost of the agents. The congress will take place next November in Maastricht, Netherlands. |
|
|
Publication accepted ICAART2012 |
|
|
|
|
Recently we have reported acceptance of an article in poster format in the next edition of the congress for editing ICAART 2012: http://www.icaart.org/ The article title is "B2DI: A BDI Agent model with Bayesian Belief updating based on causal MSBN" Alvaro Carlos Carrera and A. Iglesias. The article proposes a model that combines the already widely known BDI model for agents with distributed reasoning techniques based on multi-sectioned Bayesian networks (MSBN). The congress will take place next February in Vilamoura, Algarve, Portugal. |
|
SE4ND first beta (Simulation Environment for Network Diagnosis) |
|
|
|
|
We recently therms the first beta version of SE4ND (Simulation Environment for Network Diagnosis). SE4ND is a simulation system for evaluating network fault diagnostic agents based on MASON (http://cs.gmu.edu/ eclab ~ / projects / mason /). In this first version we have implemented a scenario that recreates the dispositovs present in a fiber optic network (FTTH - Fiber To The Home). An agent is responsible for generating random failures in the devices, which can be observed through a graphical interface. In future iterations, add other scenarios (ADSL, P2P, etc..) To get simulate different scenarios and evaluate complex and multi-agent systems in different stress situations depending on various factors such as response time, ability to solve diagnostic autonomy. |
|
Start a project with Telefónica I + D |
|
|
|
|
In October we started a new project with Telefónica I + D for diagnostics automation of network management. This project is of great interest because of the technologies involved, from web scraping to maintain multi-agent systems as a central block of the uncertainty management, a feature present at all times in any diagnostic process. The scenario in which you work on this project is broadband access to end users. The objective of the system is to provide a rapid diagnosis for operators who process incidences of users without having to manually launch all possible evidence. This will speed up considerably the average time of diagnosis and in impacting customer satisfaction of the operator. The result of the diagnostic process the operator is displayed in a list of hypotheses about possible failures that may be occurring at that moment with that user. Each hypothesis is associated a percentage that represents the confidence you have on that particular fault. This is achieved through probabilistic techniques used in the Bayesian inference engine to crowd all the available information the user and using heuristics to manage the data that are unknown at that time. The final result of the project will generate a system to be implemented in a central node in the network of support to end users "Telefónica Spain" located in Valladolid. |
|
Call - Reading PFC - Agents with Bayesian Reasoning for Network Management |
|
|
|
Title: Bayesian Reasoning Module for a BDI agent architecture. Application to diagnosis in FTTH networks. Author: Jesus Lopez Tutor: Álvaro Race Location: B-225 Day: Friday September 30 Time: 10:00 AM Summary. This project aims to develop an infrastructure of distributed reasoning for multiagent systems. In particular, the project has focused on fault diagnosis in telecommunication networks. Given the uncertainty in the diagnosis and distribution of data, this project has continued a previous project that proposed the use of Bayesian networks to manage this uncertainty.
In this project, we address one of its limitations: the inability to manage a single Bayesian network in large networks. To this end, we have worked on the study of multi-section Bayesian networks, which facilitate the distribution of a Bayesian network at various nodes.
The project has proposed a mechanism for communication between nodes to maintain consistency, which has been implemented by JGroups middleware multicasting. To validate this model, the application has been developed on a stage at diagnosis FFTH, integrating the system of reasoning in a multiagent platform that unfolds on stage simulated FTTH. |
|
On Monday (October 3, 2011) will be held from 12:00 to 13:00 in room B-225 MASON introductory class. MASON is a library of multi-agent simulation based on discrete event implementers entirely in Java. It is designed to be the basis for any purpose simulations and also provides enough functionality that may be required in light and fast simulations. MASON contains both libraries also modeled as 2D and 3D visualization. Although it is not necessary to perform pre-register, send an email to a.carrera (at) gsi.dit.upm.es to estimate attendance. |
|
On Monday (October 3, 2011) will be held from 12:00 to 13:00 in the laboratory B-225 MASON introductory class. MASON is a framework of multi-agent simulation designed to be the basis of complex simulations, and also provides functionality to perform simulations in case of light needed. It is written entirely in Java and contains libraries of 2D and 3D user-friendly. Although it is not necessary to perform pre-register, send an email to a.carrera (at) gsi.dit.upm.es to estimate attendance. |
|
|
|
|
<< Start < Prev 1 2 3 4 5 6 7 8 Next > End >>
|
|
Page 4 of 8 |
Home 
RSS