Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2020-09-28 Cooperative journals: 《计算机应用研究》
Abstract： Image classification techniques based on deep neural networks have achieved great success in recent years. However, recent studies have shown that deep neural network are vulnerable to the attack of adversarial examples. To solve this problem, some works train networks by adding Gaussian noise to the image. Thereby improving the ability of the network to defend adversarial examples, but the method does not consider the sensitivity of the network to different areas in the image when adding noise. To solve this problem, this paper proposed an adversarial training algorithm based on gradient guidance noise addition. When training the network, adding adaptive noise to different areas based on the sensitivity, adding large noise to the more sensitive areas, suppressing the sensitivity of the network to image changes, adding less noise to the less sensitive areas and improves the network classification accuracy. Compared with the existing algorithms on the cifar-10 dataset, the experimental results show that the proposed method effectively improved the accuracy of neural networks when classifying adversarial examples.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-04-01 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at the problem that the existing network does not consider the heterogeneity of mobile network nodes and the failure to construct a destructive virus propagation model, this paper proposed a destructive virus propagation model based on heterogeneous mobile networks. By considering the heterogeneity of mobile network nodes, this paper further divided the uninfected state into a new system state and an old system state, and combined with the latent and explosive characteristics of the destructive virus, divided the infection state into a latent state and an explosive state. This paper calculated the equilibrium point and propagation threshold of the model, and pointed out that when the propagation threshold is greater than 1, the model is unstable at the positive equilibrium point; when the propagation threshold is less than 1, the model is locally asymptotically stable at the positive equilibrium point. This paper carried out the simulation experiments on the NW small world network and the BA scale-free network. The simulation results showed that the virus propagation speeds of the two networks are different, and the NW network has a complete elimination of the virus, but BA does not have this situation.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-11-29 Cooperative journals: 《计算机应用研究》
Abstract： The traditional method of calculating the similarity of the Microblog users based on the user's blog content and the number of common friends has the problem of excessive potential error, and the similarity calculation model based on the user's multi-source background information has high computational complexity and ignore the user's interest and other issues, the author puts forward a combined with user's interest and background information to calculate the comprehensive similarity (BIBS) . The method extracts the user's interest from the user's tag. When the user's tag is missing, the user's interest is indirectly obtained by clustering the important user in the user's attention network, and the user's interest similarity is calculated, and then the background similarity of the user is calculated according to the background information such as the gender, age and location of the user, so that the most similar users are hierarchically mined. Experiments and analysis based on the data of Sina Microblog show that compared with MISUR algorithm based on the similarity of multi-source information, the algorithm can improve the accuracy, recall rate and F-measure by 8.1%, 16.7% and 13.6% respectively with less time consuming, which proves the effectiveness and accuracy of the proposed BIBS method.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-24 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at the problem of ignores the user's interest in the key words and the data sparseness in traditional collaborative filtering algorithm. We proposed a collaborative filtering recommendation algorithm combined with the user interest degree clustering. We using user ratings for projects and extracting keywords from item attributes. A new Rating Frequency-Inverse Item Frequency algorithm is proposed. According to the target users' scoring frequency for a key word and the frequency of the keyword being evaluated by all users. We get users' preferences for keywords, form user preference matrix, and cluster on the basis of this matrix. Then we use logistic function to get users' interest in projects. Clear user preferences and find similar users of target users in the clusters. Then extract N items from neighbors' preferences, and recommend users. Experimental results show that the algorithm accuracy rate is always better than the traditional algorithm. it, s more accurate to judge the user interest, alleviating the problem of data sparseness, and effectively improves the accuracy and efficiency of recommendation.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-18 Cooperative journals: 《计算机应用研究》
Abstract： In order to study the role of opinion leaders in the dissemination of Sina micro-blog and the propagation mode of micro-blog's life cycle in social networks effectively, the paper proposed a propagation prediction OLB model. Firstly, we crawled the micro-blog data, then analyzed the data; Secondly, we fitted a mathematical expression of the four factors related to influence, and gave the weights calculation method through the analytic hierarchy process; Finally, we constructed the OLB model by using the influence of computation and the relationship between forwarding number and related factors. We analyzed the communication role of opinion leaders and micro-blog life cycle through the experiment. The simulation results showed that the influence of opinion leaders was positively proportional to the spread of micro-blog in micro-blog information dissemination. The average error of the four sets of data were 1.0%, 5.0%, 2.4% and 5.1% respectively by error analysis. The OLB model is reasonable and effective for predicting micro-blog propagation patterns.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-24 Cooperative journals: 《计算机应用研究》
Abstract： In order to study the impact of differences in user communication capabilities on the spread of spam. This paper divided the population in the communication network into four compartments I(Ignorant) , W(Weakly spreader) , S(Strongly spreader) , R(Removal) and established state transition rules according to the theory of compartment modeling. Using differential dynamical system theory, this paper developed an IWSR spam propagation model with different communication capabilities. By calculating the basic reproduction number of the model and using the Hurwitz criterion, Lyapunov stability theorem and LaSalle invariance principle, this paper obtained the conditions for local stability and global stability of the spam information propagation equilibrium points. Finally, the comparison experiment between the numerical simulation and the Sina Weibo dataset proves that the conclusion is reasonable and effective.
Hits
Hits
Downloads
Comment