Abstract Differential privacy (DP) provides formal guarantees that the output of a database query does not reveal too much information about any individual present in the database. While many differentially private algorithms have been proposed in the scientific literature, there are only a few end-to-end implementations of differentially private query engines. Crucially, existing systems assume that each individual is associated with at most one database record, which is unrealistic in practice. We propose a generic and scalable method to perform differentially private aggregations on databases, even when individuals can each be associated with arbitrarily many rows. We express this method as an operator in relational algebra, and implement it in an SQL engine. To validate this system, we test the utility of typical queries on industry benchmarks, and verify its correctness with a stochastic test framework we developed. We highlight the promises and pitfalls learned when deploying such a system in practice, and we publish its core components as open-source software.

We introduce the concept of local bucket error for the mini-bucket heuristics and show how it can be used to improve the power of AND/OR search for combinatorial optimization tasks in graphical models (e.g. MAP/MPE or weighted CSPs). The local bucket error illuminates how the heuristic errors are distributed in the search space, guided by the mini-bucket heuristic. We present and analyze methods for compiling the local bucket-errors (exactly and approximately) and show that they can be used to yield an effective tool for balancing look-ahead overhead during search. This can be especially instrumental when memory is restricted, accommodating the generation of only weak compiled heuristics. We illustrate the impact of the proposed schemes in an extensive empirical evaluation for both finding exact solutions and anytime suboptimal solutions.

Best-first search can be regarded as anytime scheme for producing lower bounds on the optimal solution, a characteristic that is mostly overlooked. We explore this topic in the context of AND/OR best-first search, guided by the MBE heuristic, when solving graphical models. In that context, the impact of the secondary heuristic for subproblem ordering may be significant, especially in the anytime context. Indeed, our paper illustrates this, showing that the new concept of bucket errors can advise in providing effective subproblem orderings in AND/OR search.

The paper explores the potential of look-ahead methods within the context of AND/OR search in graphical models using the Mini-Bucket heuristic for combinatorial optimization tasks (e.g., weighted CSPS or MAP inference). We study how these methods can be used to compensate for the approximation error of the initially generated Mini-Bucket heuristics, within the context of anytime Branch-And-Bound search.

We explore the use of iterative cost-shifting as a dynamic heuristic generator for solving MPE in graphical models via Branch and Bound. When mini-bucket elimination is limited by its memory budget, it may not provide good heuristics. This can happen often when the graphical model has a very high induced width with large variable domain sizes. In addition, we explore a hybrid setup where both MBE and the iterative cost-shifting bound are used in a combined heuristic. We compare these approaches with the most advanced statically generated heuristics.