Research On Multi-objective Sequential Three-way Decision Malware Detection Model

In recent years,the rapid growth of information and networks has facilitated economic and social development and advancement,while also posing new challenges to network security.The exponential increase in types and quantities of malware presents a significant threat to computer security and user privacy.In response to this problem,many researchers have conducted research on malware detection.However,existing malware detection methods are typically based on the two-way decision,and suffer from the problem of insufficient feature extraction,resulting in high rates of false positives and false negatives.To address this issue,this paper introduces sequential three-way decision into malware detection,and constructs the malware detection models.Furthermore,we improve the many-objective optimization algorithm from the perspective of algorithms to solve the model.The research contents of this paper are as follows:In order to alleviate the problem of high false positive and false negative rates caused by insufficient feature extraction in malware detection,this paper introduces sequential three-way decision and designs a multi-objective sequential three-way decision malware detection model.First,the features of malware data are extracted,and a multi-granularity feature set is constructed.Secondly,the sequential three-way decision method is introduced to make three-way decisions on the data at different granularities in sequence.On this basis,the threshold acquisition problem in the classic sequential three-way decision model is transformed into a multi-objective optimization problem,and a multi-objective sequential three-way decision model is designed,which is solved using multi-objective optimization algorithms.Finally,experimental results demonstrate the effectiveness of introducing sequential three-way decision into malware detection,and also verify that using multi-objective optimization algorithms to optimize the three-way decision threshold can effectively improve the overall performance of malware detection.On the basis of the multi-objective sequential three-way decision malware detection model,considering the decision risk cost generated during the detection process and in order to further improve the performance of the malware detection model,this paper proposes a many-objective sequential three-way decision malware detection model.The model optimizes the comprehensive classification performance,decision efficiency,and decision risk cost in the sequential process of the malware detection model,and uses the many-objective optimization algorithm to solve the proposed model.The experiments demonstrate the effectiveness of the constructed many-objective model in malware detection.To enhance the efficiency of model solving process,this paper presents an advanced many-objective optimization algorithm Addressing the problem of selection pressure loss caused by an increase in target dimensions,this paper proposes a many-objective optimization algorithm based on a hybrid selection strategy,starting from the aspect of environmental selection.The strategy concurrently integrates multiple selection operators to select better individuals for the next generation population,thus improving the overall evolutionary performance of the algorithm.Simulation experimental results demonstrate that the proposed algorithm performs well on standard test sets,and exhibits algorithmic advantages when applied to solve the many-objective sequential three-way decision malware detection model.