Software forced safety protection method based on external environment model learning

Abstract

The invention discloses a software forced safety protection method based on external environment model learning, which comprises the following steps: for an external computer system, collecting a historical operation log of the external computer system and matching an execution result of a software operation sequence queried by parts in an MAT framework, calculating the execution result according to a preset first model learning algorithm to obtain a Mealy machine model, analyzing the Mealy machine model in combination with a preset protocol to obtain a program path violating the protocol, and generating a forcifier according to the program path violating the protocol; for the external environment, obtaining logs of the internal system, extracting an interaction event sequence of the external environment and the internal system as an observation sequence, calculating the observation sequence according to a preset second model learning algorithm to obtain a probability model, and generating a monitor through the probability model. According to the method, the learning process is combined with the log, so that the learning efficiency is greatly improved.

Description

technical field [0001] The invention relates to the field of computer systems, in particular to a software mandatory security protection method based on external environment model learning. Background technique [0002] As the scale of the software system becomes larger and the logic becomes more and more complex, the environment of the software system becomes more and more changeable, and the objects interacting with the software system become more and more diverse. In addition, it is difficult for software developers to consider at the beginning of development. If it is comprehensive, then more and more software security problems will naturally be exposed. In the face of these security problems, the main method adopted at this stage is to passively compensate for software security risks based on experience. Powerful vulnerability mining tools and methods are needed to discover and fix vulnerabilities as early as possible, but in fact, due to the huge scale of the software...

AI Technical Summary

Problems solved by technology

Finally, when applying to solve actual cases, we will face a problem. The learning process is an active learning process. The member query operation needs to execute the learned system to obtain the output corresponding to the input, and this process takes time due to the software execution. The learning process requires a large number of member query operations, making the learning process take a lot of time

Images