@article {Benkaouz201648, title = {Nearest neighbors graph construction: Peer sampling to the rescue}, journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}, volume = {9944 LNCS}, year = {2016}, note = {cited By 0}, pages = {48-62}, abstract = {In this paper, we propose an efficient KNN service, called KPS (KNN-Peer-Sampling). The KPS service can be used in various contexts e.g. recommendation systems, information retrieval and data mining. KPS borrows concepts from P2P gossip-based clustering protocols to provide a localized and efficient KNN computation in large-scale systems. KPS is a sampling-based iterative approach, combining randomness, to provide serendipity and avoid local minimum, and clustering, to ensure fast convergence. We compare KPS against the state of the art KNN centralized computation algorithm NNDescent, on multiple datasets. The experiments confirm the efficiency of KPS over NNDescent: KPS improves significantly on the computational cost while converging quickly to a close to optimal KNN graph. For instance, the cost, expressed in number of pairwise similarity computations, is reduced by ≈23\% and ≈49\% to construct high quality KNN graphs for Jester and MovieLens datasets, respectively. In addition, the randomized nature of KPS ensures eventual convergence, not always achieved with NNDescent. {\textcopyright} Springer International Publishing AG 2016.}, doi = {10.1007/978-3-319-46140-3_4}, url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84990068831\&doi=10.1007\%2f978-3-319-46140-3_4\&partnerID=40\&md5=001c7ac565106835cc5cd7a01e01a3c4}, author = {Benkaouz, Y.a and Erradi, M.a and Kermarrec, A.-M.b} } @conference {Benkaouz201672, title = {Work in progress: K-nearest neighbors techniques for ABAC policies clustering}, booktitle = {ABAC 2016 - Proceedings of the 2016 ACM International Workshop on Attribute Based Access Control, co-located with CODASPY 2016}, year = {2016}, note = {cited By 0}, pages = {72-75}, abstract = {In this paper, we present an approach based on the K- Nearest Neighbors algorithms for policies clustering that aims to reduce the ABAC policies dimensionality for high scale systems. Since ABAC considers a very large set of attributes for access decisions, it turns out that using such model for large scale systems might be very complicated. To date, researchers have proposed to use data mining techniques to discover roles for RBAC system construction. In this work in progress, we consider the usage of KNN-based techniques for the classification of ABAC policies based on similarity computations of rules in order to enhance the ABAC exibility and to reduce the number of policy rules. {\textcopyright} 2016 ACM.}, doi = {10.1145/2875491.2875497}, url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84966549589\&doi=10.1145\%2f2875491.2875497\&partnerID=40\&md5=2378f9f3208a82bfc833b26f1a7ef267}, author = {Benkaouz, Y.a and Erradi, M.a and Freisleben, B.b} } @article {Benkaouz201594, title = {Distributed privacy-preserving data aggregation via anonymization}, journal = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}, volume = {9466}, year = {2015}, note = {cited By 0}, pages = {94-108}, abstract = {Data aggregation is a key element in many applications that draw insights from data analytics, such as medical research, smart metering, recommendation systems and real-time marketing. In general, data is gathered from several sources, processed, and publicly released for data analysis. Since the considered data might contain personal and sensitive information, special handling of private data is required. In this paper, we present a novel distributed privacy-preserving data aggregation protocol, called ADiPA. It relies on anonymization techniques for protecting personal data, such as k-anonymity, l-diversity and t-closeness. Its purpose is to allow a set of entities to derive aggregate results from data tables that are partitioned across these entities in a fully decentralized manner while preserving the privacy of their individual sensitive inputs. ADiPA neither relies on a trusted third party nor on cryptographic techniques. The protocol performs accurate aggregation when communication links and nodes do not fail. {\textcopyright} Springer International Publishing Switzerland 2015.}, doi = {10.1007/978-3-319-26850-7_7}, url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84961141090\&doi=10.1007\%2f978-3-319-26850-7_7\&partnerID=40\&md5=518bf129894b27ee29b635e53f9baf0c}, author = {Benkaouz, Y.a and Erradi, M.a and Freisleben, B.b} }