Two methods for privacy preserving data mining with malicious participants

k-Anonymous data collection

“…Then she sends the encryptions to the miner.For example, suppose there are three respondents 1, 2, and 3, having data ðM; 23; strokeÞðF; 24; fluÞðM; 23; allergyÞ, respectively, where the first two fields of each piece of data are the quasi-identifier. After proper encoding, respondent 1's data becomes (123, 55), respondent 2's data becomes (224, 38), and respondent 3's data becomes (123,49), where the quasi-identifiers are 123, 224, and 123, respectively. Then the miner receives an encryption of 123 and an encryption of 55 from respondent 1; the public key used to encrypt them is the product of the miner's public key and the DCH's public key.…”

Service reliability modeling of distributed computing systems with virus epidemics

“…In the literature of reliability research, many studies have been devoted to intentional attack protections by designing protection strategies for different systems (e.g. power substations, defense systems, etc) [16][17][18][19][20], but few have investigated the attacks with epidemic characteristics, such as computer virus [21]. In the field of epidemiology modeling, some research works have addressed the virus spreading issue in computer networks, but the emphasis is on the speed and range of the spreading [21][22][23].…”

Learning latent variable models from distributed and abstracted data

“…Concerns including bandwidth limitation and data privacy [39,28,19,38] restrict the distributed data from being pooled together for analysis, forming the causes for the challenge of carrying out global knowledge discovery from distributed data. This paper describes an attempt to address the underlying issues using a data abstraction approach.…”