Robust Influence Maximization

Abstract

In this paper, we address the important issue of uncertainty in the edge influence probability estimates for the well studied influence maximization problem -the task of finding k seed nodes in a social network to maximize the influence spread. We propose the problem of robust influence maximization, which maximizes the worst-case ratio between the influence spread of the chosen seed set and the optimal seed set, given the uncertainty of the parameter input. We design an algorithm that solves this problem with a solutiondependent bound. We further study uniform sampling and adaptive sampling methods to effectively reduce the uncertainty on parameters and improve the robustness of the influence maximization task. Our empirical results show that parameter uncertainty may greatly affect influence maximization performance and prior studies that learned influence probabilities could lead to poor performance in robust influence maximization due to relatively large uncertainty in parameter estimates, and information cascade based adaptive sampling method may be an effective way to improve the robustness of influence maximization.

Robust Densest Subgraph Discovery

“…Recently, the theory of robust optimization has been widely applied to tasks in knowledge discovery and data mining, particularly to graph mining tasks. For example, Chen et al [14] and He and Kempe [22] studied robust influence maximization, which is a robust variation of the popular graph mining task called influence maximization. Their focus was on the influence maximization counterpart of our work; they aimed to find a subset of vertices that exhibits a large robust ratio in terms of the influence.…”

Influence maximisation beyond organisational boundaries

“…Moreover, some work, such as [16] or [17], uses robust optimisation techniques. In [16], the authors address the uncertainty in the edge influence probability.…”

“…In [16], the authors address the uncertainty in the edge influence probability. More precisely, in the model, instead of the exact probability of influence, every edge of the social graph is associated with some interval in which the true probability may lie.…”

The evaluation and application of node influence in dynamic networks based on evolving communities

“…[1][2][3][4] In a cellular network where mobile worms exist, if we can patch an optimal set of influential phones, the propagation of worms can be effectively prevented. [1][2][3][4] In a cellular network where mobile worms exist, if we can patch an optimal set of influential phones, the propagation of worms can be effectively prevented.…”