Search

Deep learning struggles with unsupervised tasks like community detection in networks. This work proposes the Enhanced Community Detection with Structural Information VGAE (VGAE-ECF) method, a method that enhances variational graph autoencoders (VGAEs) for community detection in large networks. It incorporates community structure information and edge weights alongside traditional network data. This combined input leads to improved latent representations for community identification via K-means clustering. We perform experiments and show that our method works better than previous approaches of community-aware VGAEs. [ABSTRACT FROM AUTHOR]

The branching-time transformation is a recent technique for optimizing Chain Datalog programs. In this paper we propose a significant extension of the branching-time transformation which we believe opens up a promising new direction of research in the area of value-propagating Datalog optimizations. More specifically, the proposed transformation can handle more general programs that allow multiple consumptive occurrences of variables in the bodies of clauses. This extension is achieved by using as target language the temporal logic programming formalism Datalog_{nS} enriched with choice predicates (a non-deterministic construct that was originally introduced in the area of intensional logic programming). We demonstrate the correctness of the transformation and propose several optimizations that can be applied to the target code. Moreover, we define a bottom-up proof procedure that applies to the target programs and demonstrate that it always terminates (despite the fact that the Herbrand base of these programs is generally infinite). [ABSTRACT FROM AUTHOR]