Game strategy-based dynamic flow graph online balanced division method and storage medium

Abstract

The invention relates to a game strategy-based dynamic flow graph online balanced division method and a storage medium. The online balanced division method comprises the following steps of: setting a target division block obtained at a previous moment as a game participant; for all game participants, setting a strategy set of the participants based on all possible actions of the newly added vertexes; constructing a dynamic flow graph online equilibrium division game model based on the equilibrium degree ED and the modularity MD; calculating utility values of each participant under all strategies by using a dynamic flow graph online balanced division game model; solving the online equilibrium division model to obtain a Nash equilibrium solution; updating the target block according to the strategy of each participant corresponding to the Nash equilibrium solution; and completing online balanced division of the dynamic flow graph. Compared with the prior art, the method has the advantages of more reasonable block division, high division speed, dynamic property, incremental property, high reusability and the like.

Description

technical field [0001] The invention relates to the technical field of dynamic flow graph division, in particular to an online balanced division method and storage medium of dynamic flow graphs based on a game strategy. Background technique [0002] With the development of the Internet of Things, 5G and Industry 4.0, we are currently in the era of achieving a higher level of artificial intelligence. Application fields such as smart transportation, smart wearable devices, smart home, and social networks continue to generate exponentially growing, interrelated graph data over time, and continue to arrive in a "stream" manner, thus forming a gradually increasing scale The flow graph of big data is used as the basis for machine learning and artificial intelligence. For example, in the field of intelligent transportation, driverless cars continuously obtain road condition information through radar and visual sensors to make timely decisions; in the global ocean observation syste...

AI Technical Summary

Problems solved by technology

Since all vertex and edge graph data need to be stored, the full partition method is not suitable for large-scale graph data, and it is even more difficult to partition dynamic flow graphs with high real-time requirements.

The incremental partitioning method avoids redundant calculations to a certain extent and improves partitioning efficiency. However, this method is sensitive to the scale of local changes in dynamic graphs. When the graph changes beyond a certain scale, its time complexity will approach that of the full partitioning method.

Streaming partitioning methods, such as Fennel and WStream, aim at fast online partitioning. On the basis of the existing partitioning results, the newly added flow graph data that arrives continuously is divided. This type of method is suitable for large Scale flow graph data, but the division process does not consider the competition and cooperation relationship between the target division blocks, and it is difficult to find the best division strategy for the allocation and division of new vertices

Images