Abstract | ||
---|---|---|
We study the problem of a user who has both public and private data, and wants to release the public data, e.g. to a recommendation service, yet simultaneously wants to protect his private data from being inferred via big data analytics. This problem has previously been formulated as a convex optimization problem with linear constraints where the objective is to minimize the mutual information between the private and released data. This attractive formulation faces a challenge in practice because when the underlying alphabet of the user profile is large, there are too many potential ways to distort the original profile. We address this fundamental scalability challenge. We propose to generate sparse privacy-preserving mappings by recasting the problem as a sequence of linear programs and solving each of these incrementally using an adaptation of Dantzig-Wolfe decomposition. We evaluate our approach on several datasets and demonstrate that nearly optimal privacy-preserving mappings can be learned quickly even at scale. |
Year | Venue | Field |
---|---|---|
2014 | UNCERTAINTY IN ARTIFICIAL INTELLIGENCE | Data mining,User profile,Computer science,Theoretical computer science,Artificial intelligence,Mutual information,Convex optimization,Big data,Machine learning,Alphabet,Scalability |
DocType | Citations | PageRank |
Conference | 2 | 0.46 |
References | Authors | |
15 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Salman Salamatian | 1 | 36 | 5.69 |
Nadia Fawaz | 2 | 215 | 16.58 |
Branislav Kveton | 3 | 455 | 49.32 |
Nina Taft | 4 | 2109 | 154.92 |