Intrinsic Node Resistance and Its Application in Depreciatory Adversarial Attacks

ABSTRACT Edge‐weighted graphs are broadly utilized in the analysis of path‐finding and optimization algorithms. Yet when it comes to network robustness, little attention is given to the properties of vertices. In this paper, we introduce a novel network model where each node is assigned an intrinsic resistance parameter from a uniform distribution. Introduction of such nodal parameters aids us in measuring overall network resilience. We also generalize an existing model governing network robustness measurements known as the Effective Node Resistance (abbreviated henceforth as ENR) to networks containing vertices with internal resistances. We will later provide state‐of‐art definitions for network attacker systems, specifically a generalization of these systems we call the Depreciatory Attacker Systems. Finally, the effectiveness of our proposed model in comparison to previous network models on defense against various types of depreciatory adversarial attacks will be analyzed through computer‐assisted simulations.