Interpretation method of global view network system based on deep learning

Abstract

The invention relates to the technical field of internet information, in particular to an interpretation method of a global view network system based on deep learning. According to the method, causalinterpretation and conversion are carried out on decisions of a computer network system based on deep learning under the global visual field condition. Firstly, an original network system is trained by adopting a deep reinforcement learning method; after original system training based on deep learning is completed, modeling is performed on a generated global configuration result in a hypergraph mode, key point-hyperedge connections in the hypergraph are analyzed, and the influence of each point-hyperedge connection on a final global configuration result is scored, so that a network administrator understands key components in decision making. According to the method, the understanding difficulty of an original global view network system based on deep learning is greatly reduced, and a network administrator can understand the decision process conveniently. When the interpretation method is deployed on an actual system, a network administrator can understand and correct the decision process of the original global view network system.

Description

technical field [0001] The invention relates to the field of Internet information technology, in particular to an interpretation method of a global vision network system based on deep learning. Background technique [0002] Computer network systems can generally be divided into systems with local vision and systems with global vision. Partial vision means that the network system is deployed on the server side, client side or middleware, switch (such as: congestion control system), these systems with local vision can only observe the information of one point in the system and make decision making. Systems with a global view include network management controllers, traffic engineering scheduling systems, software-defined network controllers, etc. These systems with a global view can observe and make decisions on multiple devices in the network. In terms of decision-making logic, some existing global vision network systems use deep learning technology as their decision-making ...

AI Technical Summary

Problems solved by technology

However, the decision-making methods of these existing deep learning-based systems cannot be understood by network administrators: neural networks often contain tens of thousands of neurons, and the final conclusion is drawn after a series of nonlinear calculations

Therefore, despite performing well in training, network administrators do not understand the logic of their decisions and are therefore often difficult to gain trust

Images