Construction And Optimization Of Multi-objective Malicious Code Detection Model

Malicious code detection and classification play a very important role for the maintenance of cyber security.Especially in recent years,with the rapid development of big data and Internet of things,various new cyber technologies emerge in an endless stream.But the development of these technologies has also contributed to the growing proliferation of malicious code.Therefore,the detection and classification of malicious code are facing more and more challenges.This article analyzes the problems in the existing detection and classification methods and constructs an efficient and accurate model of malicious code detection and classification based on deep learning and intelligent optimization algorithms.The main contents of this article are as follows:In view of the existing detection method is difficult to deal with the rapid growth of malicious code.Therefore,according to the theory of detection method based on convolution neural network limitations,the Spatial Pyramid Pooling(SPP)structure is employed to improve the neural network structure for malicious code detection.At the same time in order to reduce the influence of the sparse malicious code dataset,we put forward a kind of efficient data enhancement method combines Generative Adversarial Networks(GANs)and Convolutional neural networks(CNNs)to build enough available datasets.The test results show that the malicious code image generated by the proposed data enhancement method is more authentic and the established neural network effectively overcomes the limitation of unified input.The constructed malicious code dataset will be used for the subsequent work.After the construction of the neural network,we analyzed the existing methods of single evaluation criterion respectively.In order to further reduce the negative effects of imbalanced dataset,reasonable objective functions are designed and the multi-objective malicious code detection model is build based on neural network and NSGA-II algorithm.The experimental results show that performance of the constructed model has great improvement compared with the existing detection methods;Then,more factors are introduced to evaluate the detection model on the basis of multi-objective model.The many-objective malicious code detection model based on neural network and NSGA-III algorithm,which proved further improves the detection performance of the model.The structural adjustment of neural network is a very important problem in the construction of the above detection model.We analyze the existing structural adjustment methods and propose a method to adjust the overall structure of neural network by using optimization algorithm.Firstly,NSGA-II algorithm is utilized to adjust its structural parameters according to the characteristics of GANs.In order to verify the effectiveness of the method,a contrast experiment was designed on the malicious code and MNIST datasets respectively.After analyzing the results,we found that the adjusted network has better image generation ability.Then the structural parameters in the CNNs are analyzed and optimized by NSGA-III.Experimental results show that the adjusted network has better performance on the malicious code,MNIST and CIFAR datasets.