A Standardised Neighbourhood Interface through Massive Parallel Computing for Mining Digital Data

Abstract

Any of the elements now has some details on the global position of the device in a vast network of computers or wireless devices. Most functionality of the framework, including message delivery, information collection and load sharing, is focused on global status modulization. It was referred to each peer's load as a model of the device as the output of the function. As the state of the environment continues to evolve, the models must be kept up to date. Worldwide data mining models such as choice trees and k-means clustering can be very expensive, due both to the device size and connectivity costs, in large distributed systems. The costs escalate more if the data shifts quickly in a complex situation. In this paper, a two-stage approach to handling these costs has been presented. Second, it was defined a highly successful local algorithm to track a large variety of data mining models. This procedure is then used as a feedback circuit to control complex information structures, such as the segmentation of k-means. A detailed scientific review supports the theoretical arguments.