부산대학교 도서관

A crucial issue for machine learning on vertices is the collection of frequent subgraphs from a group of networks or a single graph. However, if the initial networks contained sensitive information, disclosing the discovered similar subset might pose major risks to people's privacy. The issue of occasional graph g minor (Spaghetti monster) is discussed in this study under a strict privacy preserving architecture. We outline a numerous private information FSM technique we refer to as Generator. In Two - dimensional array, the whole first step involved the secret discovery of widespread summary study, and the second phase involves the computation of the anarchic supporting of each recognised often snippet. We provide a commonly substrings locator, that might improve the quality of detected commonly subgraphs by option reduction, particularly to covertly share similar subgraphs. Moreover, we develop a vector noisy assist calculating method that uses the inclusion correlations between both of the discovered regular sub bands to increase the accuracy of the loud offers in order to compute the quiet follower of each detected shared subnetwork. We contend that Facts and figures satisfies more could privacy via formalised risk control. Lab tests on real study shows that Facts and figures can achieve great data value and covertly share similar subsets.