Adversary: Somebody who, whether intentionally or not, reveals sensitive, private information Adversarial model: Formal description of the unique characteristics of a particular adversary Attribute disclosure: A privacy breach wherein some descriptive attribute of somebody is revealed Identity disclosure: A privacy breach in which a presumably anonymous person is in fact identifiable k-P-anonym...

Although various methods have been proposed for pedestrian attribute recognition, most studies follow the same feature learning mechanism, \ie, a shared image to classify multiple attributes. However, this mechanism leads low-confidence predictions and non-robustness of model in inference stage. In paper, we investigate why is case. We mathematically discover that central cause optimal cannot m...

We propose a new encryption primitive called Membership Encryption. Let P(G) be a privacy-preserving token on a group attribute/identity G, such that given P(G) it is hard to know the attributes in G. In this membership encryption, if an encryption takes as input an attribute A and the token P(G) , the decryption requires holding the membership A ∈ G, i.e., A belongs to this group attribute. Me...

This paper introduces an eecient indexing organization which provides for parallel retrievals. Given a p-processor massively parallel computer, our organization allows a maximum of p retrievals to be performed concurrently and thus achieves an approximate speed up of p in the best case. The index structure is based on the multi-B-tree index structure. It is compatible with current automatic ret...

in some statistical process control applications, the quality of the product is characterized by thecombination of both correlated variable and attributes quality characteristics. in this paper, we propose anovel control scheme based on the combination of two multi-layer perceptron neural networks forsimultaneous monitoring of mean vector as well as the covariance matrix in multivariate-attribu...