Graph Neural Networks (GNNs) have achieved promising performance in various real-world applications. However, recent studies find that GNNs are vulnerable to adversarial attacks. In this paper, we study a recently-introduced realistic attack scenario on graphsgraph injection attack (GIA). In the GIA scenario, the adversary is not able to modify the existing link structure or node attributes of the input graph, instead the attack is performed by injecting adversarial nodes into it. We present an analysis on the topological vulnerability of GNNs under GIA setting, based on which we propose the Topological Defective Graph Injection Attack (TDGIA) for effective injection attacks. TDGIA first introduces the topological defective edge selection strategy to choose the original nodes for connecting with the injected ones. It then designs the smooth feature optimization objective to generate the features for the injected nodes. Extensive experiments on large-scale datasets show that TD-GIA can consistently and significantly outperform various attack baselines in attacking dozens of defense GNN models. Notably, the performance drop on target GNNs resultant from TDGIA is more than double the damage brought by the best attack solution among hundreds of submissions on KDD-CUP 2020. CCS CONCEPTS• Security and privacy → Software and application security; • Mathematics of computing → Graph algorithms.

No abstract

No abstract

Wireless sensor networks (WSNs) are data centric networks to which data aggregation is a central mechanism. Nodes in such networks are known to be of low complexity and highly constrained in energy. This requires novel distributed algorithms to data aggregation, where accuracy, complexity and energy need to be optimized in the aggregation of the raw data as well as the communication process of the aggregated data. To this end, we propose in this work a distributed data aggregation scheme based on an adaptive Auto-Regression Moving Average (ARMA) model estimation using a moving window technique and running over suitable communications protocols. In our approach, we balance the complexity of the algorithm and the accuracy of the model so as to facilitate the implementation. Subsequent analysis shows that an aggregation efficiency up to 60% can be achieved with a very fine accuracy of 0.03 degree. And simulation results confirm that this distributed algorithm provides significant energy savings (over 80%) for mass data collection applications.