Multi-Agent Systems

What is it about?

we propose a more flexible approach that employs a Multi-Agent System (MAS). We shall show that the MAS-based approach combined with DDM for data intensive applications is both desirable and beneficial in the sense a MAS is a mechanism for creating goal-oriented autonomous agents within shared environments with communication and coordination facilities. DDM benefits from this goal-oriented mechanism by implementing xiv various distributed clustering, classification, and prediction techniques. Hence, this dissertation develops a novel, Multi-Agent model for distributed classification by using different classification algorithms. For instance, an agent may request help from other agents to classify unclear instances and how this can be done locally regardless of the classification algorithms they used and where they are located. The other agent communicates his/her findings (calculated classification) to other agents who may decide whether such findings are beneficial when they are used to classify instances. The decision may result in the adjustment of the agents’ prior assumptions about each data class in what can be described as mutual collaborative classification task. A MAS model, state and behavior, and communication are all developed to facilitate information sharing among agents.

Featured Image

Photo by Carlos Muza on Unsplash

Photo by Carlos Muza on Unsplash

Why is it important?

The mutual collaborative classification task in this paper can be categorized as a general case of distributed classification in which each agent can have its own classification mechanism and can decide whether and when to request information from other agents. This dissertation conducts this mutual collaborative classification task while preserving two issues, which are required in DDM applications: covering private and non-sharable data and maintaining the distinctiveness/particularity of each agent.