Implementation of Parameter Space Search for Meta Learning in a Data-Mining Multi-agent System

Author(s):  
O. Kazik ◽  
Klara Peskova ◽  
Martin Pilat ◽  
R. Neruda
2011 ◽  
Vol 467-469 ◽  
pp. 614-619
Author(s):  
Chang Hui Yang

Choosing supplier with better quick response ability becomes more and more important. In this paper, the criterion of evaluating supplier is put forward and a method of evaluating supplier is introduced. To improve the efficiency of selecting supplier, a multi-agent system of supplier selection based on evaluating supplier is developed. Recurring to the supplier’s related data collected by data-mining agent from external web-server, the weights of criteria can be confirmed. And using the system, the supplier with better QRA can be selected based on the measuring results.


Author(s):  
Nadjib Mesbahi ◽  
Okba Kazar ◽  
Saber Benharzallah ◽  
Merouane Zoubeidi ◽  
Samir Bourekkache

Today the enterprise resource planning (ERP) became a tool that enables uniform and consistent management of information system (IS) of the company with a large single database. In addition, Data Mining is a technology whose purpose is to promote information and knowledge extraction from a large database. In this paper, an agent-based multi-layered approach for data mining based k-Means through the ERP to extract hidden knowledge in the ERP database is proposed. To achieve this, the authors call the paradigm of multi-agent system to distribute the complexity of several autonomous entities called agents, whose goal is to group records or observations on similar objects classes using the k-means technique that is dedicated the task of clustering. This will help business decision-makers to take good decisions and provide a very good response time by the use of multi-agent system. To implement the proposed architecture, it is more convenient to use the JADE platform while providing a complete set of services and agents comply with the specifications FIPA.


Author(s):  
Imane Chakour ◽  
Yousef El Mourabit ◽  
Mohamed Baslam

Recently, data mining and intelligent agents have emerged as two domains with tremendous potential for research. The capacity of agents to learn from their experience complements the data mining process. This chapter aims to study a multi-agent system that evaluates the performance of three well-known data mining algorithms—artificial neural network (ANN), support vector machines (SVM), and logistic regression or logit model (LR)—based on breast cancer data (WBCD). Then the system aggregates the classifications of these algorithms with a controller agent to increase the accuracy of the classification using a majority vote. Extensive studies are performed to evaluate the performance of these algorithms using various differential performance metrics such as classification rate, sensitivity, and specificity using different software modules. In the end, the authors see that this system gives more autonomy and initiative in the medical diagnosis and the agent can dialogue to share their knowledge.


Sign in / Sign up

Export Citation Format

Share Document