Graph adversarial sample generation method by adding false nodes based on reinforcement learning

Abstract

The invention discloses a graph adversarial sample generation method by adding false nodes based on reinforcement learning. The graph adversarial sample generation method comprises the following steps: (1) obtaining the number of original graphs and a graph node classification model, and constructing a training set and a test set; adding false nodes into the original image data to obtain an initial adversarial sample; (2) constructing an attack model; (3) selecting an attack target in the training set; (4) inputting the current adversarial sample and the attack target into an attack model, selecting a node with the maximum evaluation value, and constructing a new adversarial sample; (5) inputting a new adversarial sample into the classification model, if a classification result is a targetresult, obtaining the adversarial sample and carrying out the next step, otherwise, skipping to the step (4); and (6) training the attack model, and testing and applying the trained attack model. According to the graph adversarial sample generation method, the adversarial sample of the graph is generated by adding the false nodes, and help can be provided for designing a more robust graph deep learning model.

Description

Technical field [0001] The invention belongs to the technical field of artificial intelligence information security, and in particular relates to a method for generating graph adversarial samples by adding false nodes based on reinforcement learning. Background technique [0002] A graph in graph theory is a graph composed of a number of given points and a line connecting two points. This graph is usually used to describe a certain relationship between certain things. A point is used to represent a thing and two points are connected. The line indicates that there is a certain relationship between the corresponding two things. The graph G in graph theory is an ordered two-tuple (V, E), where V is called the vertex set, that is, the set of all vertices in the graph, and E is called the edge set, that is, the edge between all vertices. set. Simply put, vertices represent things, and edges represent relationships between things. In addition, the Attributed Graph can be expressed a...

AI Technical Summary

Problems solved by technology

In the field of graph data, current research is to make the classification results of target nodes mislead by adding or deleting existing edge or node features.

But this method may be difficult to implement in actual scenarios. For example, in a social network, if you want to delete or add an edge between two users, you may need to obtain the login permissions of these users, but this is difficult to obtain in actual situations.

Images