Cooperation among independent multi-agents in a reliable data mining system
Engy F. Ramadan, Mohamed Ahmed Shalaby, Essam ElFakhrany · 2016
In recent years, data mining techniques has been used widely to address the problem of huge data sets stored on huge, heterogeneous data warehouses and may be located in different sites. Rapid growth of data arise in many areas like healthcare, social media, science, Internet...etc. Multi-agent technology has improved the processing of such huge data by the use of the agent technology along with data mining techniques in dealing with them. We found that processing time is minimized and accuracy is increased when merging data mining classification techniques in a multi-agent system. In this paper, we introduce a multiple learner multi-agents system (MLMAS) where each agent represents a classifier. Each classifier agent will work separately and cooperatively with other agents to gain best results, the results of classifiers are then combined either by a coordinator Agent using a weighted voting technique or according to the probability distribution of classifiers' results. This system has implemented using JADE package and WEKA classifiers.