Role mining algorithms address an important access control problem: configuring a role-based access control system. Given a direct assignment of users to permissions, role mining discovers a set of roles together with an assignment of users to roles. The results should closely agree with the direct assignment. Moreover, the roles should be understandable from the business perspective in that they reflect functional roles within the enterprise. This requires hybrid role mining methods that work with both direct assignments and business information from the enterprise. In this paper, we provide statistical measures to analyze the relevance of different kinds of business information for defining roles. We then present an approach that incorporates relevant business information into a probabilistic model with an associated algorithm for hybrid role mining. Experiments on actual enterprise data show that our algorithm yields roles that both explain the given user-permission assignments and are meaningful from the business perspective.

No abstract

We reconsider the toy model studied in [1] of a spatially closed Friedmann-Lemaître universe, driven by a massive scalar field, which deflates quasi-exponentially, bounces and then enters a period of standard inflation. We find that the equations for the matter density perturbations can be solved analytically, at least at lowest order in some "slow-roll" parameter. We can therefore give, in that limit, the explicit spectrum of the post-bounce perturbations in terms of their pre-bounce initial spectrum. Our result is twofold. If the pre-bounce growing and decaying modes are of the same order of magnitude at the bounce, then the spectrum of the pre-bounce growing modes is carried over to the post-bounce decaying modes ("mode inversion"). On the other hand, if, more likely, the pre-bounce growing modes dominate, then resolution at next order indicates that their spectrum is nicely carried over, with reduced amplitude, to the post-bounce growing modes.1

Multi-label classification assigns a data item to one or several classes. This problem of multiple labels arises in fields like acoustic and visual scene analysis, news reports and medical diagnosis. In a generative framework, data with multiple labels can be interpreted as additive mixtures of emissions of the individual sources. We propose a deconvolution approach to estimate the individual contributions of each source to a given data item. Similarly, the distributions of multi-label data are computed based on the source distributions. In experiments with synthetic data, the novel approach is compared to existing models and yields more accurate parameter estimates, higher classification accuracy and ameliorated generalization to previously unseen label sets. These improvements are most pronounced on small training data sets. Also on real world acoustic data, the algorithm outperforms other generative models, in particular on small training data sets.