Finding critical edges in networks through deep reinforcement learning

The network is a powerful tool to study the interaction of system units, and the edge is an important part of the network as it represents relationships between nodes. Critical edges play an irreplaceable role in information transmission between nodes and maintaining network connectivity and integrity. Therefore, the identification of critical edges in networks is an indispensable part of network analysis, which has great practical significance. Here, we propose an algorithm IKEoN. Based on the Deep Q-learning algorithm, this algorithm identify the critical edges in the network. Instead of using labeled data sets, IKEoN uses the constant interaction between agents and the environment to train the model, which reduces the influence of network noise and improves the recognition performance. The experimental results show that the proposed method outperforms the existing methods.