Differentially Private Data Release Via Wavelet Transforms

Differential Privacy (DP) has attracted considerable interest in recent years. Among the existing privacy models, epsilon-differential privacy offered the strongest privacy guarantees. According to the academic papers in the recent years, most of the existing solutions that ensure.-differential privacy can be generally divided into two distinct models. The first is the interactive model, where the data miner is only allowed to pose aggregate queries to the database, such as [7]. Another is the non-interactive model, which release a differentially private dataset and the data miner can run any data mining algorithm on the published data. DiffGen [1] is the representative of non-interactive model. In this paper, we proposed a new algorithm W-DiffGen based on the wavelet transforms, which performs better.