Minimizing Immune Costs in Social Networks Through Reinforcement Learning

This work explores the use of social relationships to control spread of information. Spread of information through various forms in social networks. Here, the authors analyze the social relationship between individuals to model of the spread of information. In other words, the authors try to capture the interaction pattern of human beings using the social contact information and investigate its impact on the spread of information. Particularly, the authors investigate the problem of minimizing the control cost of infected persons by controlling a small fraction of the population. Current spread of information research mainly focuses on formulating immunization strategies before spread of information, while ignoring the key factor of immunization cost. To address the issue of the minimizing influence in the process of dynamic spread of information. In this paper, the authors use the Independent Cascade (IC) model for simulating the random spread of real information. Combining the Reinforcement Learning (RL) method to obtain the optimal immune strategy of each state in the process of spreading. Finally, based on experiments and evaluations on 4 real-world datasets, the simulation results validate the superiority of our strategy over existing ones. As far as we know, this is the first attempt to use the RL method to solve the problem of minimizing immune costs.