In order to solve the problem of feature information loss caused by the introduction of a large number of pooling layers in traditional convolutional neural networks, based on the feature of Capsule Network (CapsNet)——using vector neurons to save feature space information, a network model 3DSPNCapsNet (3D Small Pooling No dense Capsule Network) was proposed for recognizing 3D models. Using the new network structure, more representative features were extracted while the model complexity was reduced. And based on Dynamic Routing (DR) algorithm, Dynamic Routing-based algorithm with Length information (DRL) algorithm was proposed to optimize the iterative calculation process of capsule weights. Experimental results on ModelNet10 show that compared with 3DCapsNet (3D Capsule Network) and VoxNet, the proposed network achieves better recognition results, and has the average recognition accuracy on the original test set reached 95%. At the same time, the recognition ability of the network for the rotation 3D models was verified. After the rotation training set is appropriately extended, the average recognition rate of the proposed network for rotation models of different angles reaches 81%. The experimental results show that 3DSPNCapsNet has a good ability to recognize 3D models and their rotations.

In order to prolong the lifespan of heterogeneous directional sensor network, a node scheduling algorithm based on Enhanced Coral Reef Optimization algorithm (ECRO) and with different monitoring requirements for different monitoring targets was proposed. ECRO was utilized to divide the sensor set into multiple sets satisfying the coverage requirements, so that the network lifespan was able to be prolonged by the scheduling among sets. The improvement of Coral Reef Optimization algorithm (CRO) was reflected in four aspects. Firstly, the migration operation in biogeography-based optimization algorithm was introduced into the brooding of coral reef to preserve the excellent solutions of the original population. Secondly, the differential mutation operator with chaotic parameter was adopted in brooding to enhance the optimization ability of the offspring. Thirdly, a random reverse learning strategy were performed on the worst individual of population in order to improve the diversity of population. Forthly, by combining CRO and simulated annealing algorithm, the local searching capability of algorithm was increased. Extensive simulation experiments on both numerical benchmark functions and node scheduling were conducted. The results of numerical test show that, compared with genetic algorithm, simulated annealing algorithm, differential evolution algorithm and the improved differential evolution algorithm, ECRO has better optimization ability. The results of sensor network node scheduling show that, compared with greedy algorithm, the Learning Automata Differential Evolution (LADE) algorithm, the original CRO, ECRO has the network lifespan improved by 53.8%, 19.0% and 26.6% respectively, which demonstrates the effectiveness of the proposed algorithm.

To solve the sparse and cold start problems of recommendation system, a Trust Regularization Euclidean Embedding (TREE) algorithm by fusing trust information was proposed. Firstly, the Euclidean embedding model was employed to embed the user and project in the unified low-dimensional space. Secondly, to measure the trust information, both the project participation degree and user common scoring factor were brought into the user similarity calculation formula. Finally, a regularization term of social trust relationship was added to the Euclidean embedding model, and trust users with different preferences were used to constrain the location vectors of users and generate the recommendation results. In the experiments, the proposed TREE algorithm was compared with the Probabilistic Matrix Factorization (PMF), Social Regularization (SoReg), Social Matrix Factorization (SocialMF) and Recommend with Social Trust Ensemble (RSTE) algorithms. When dimensions are 5 and 10, TREE algorithm has the Root Mean Squared Error (RMSE) decreased by 1.60% and 5.03% respectively compared with the optimal algorithm RSTE on the dataset Filmtrust.While on the dataset Epinions, the RMSE of TREE algorithm was respectively 1.12% and 1.29% lower than that of the optimal algorithm SocialMF. Experimental results show that TREE algorithm further alleviate the sparse and cold start problems and improves the accuracy of scoring prediction.

Plant recognition includes two kinds of tasks:specimen recognition and real-environment recognition. Due to the existence of background noise, real-environment plant image recognition is more difficult. To reduce the weight of Convolutional Neural Networks (CNN), to improve over-fitting, to improve the recognition rate and generalization ability, a method of plant identification with Family Priority (FP) was proposed. Combined with the lightweight CNN MobileNet model, a plant recognition model Family Priority MobileNet (FP-MobileNet) was established by means of migration learning. On the single background plant dataset flavia, the MobileNet model achieved 99.8% of accuracy. For the more challenging real-environment flower dataset flower102, when the number of samples in the training set was greater than that in the test set FP-MobileNet achieved 99.56% of accuracy. When the number of samples in the training set was smaller than that in the test set, FP-MobileNet still obtained 95.56% of accuracy. The experimental results show that the accuracies of FP-MobileNet under two different data set partitioning schemes are both higher than those of the pure MobileNet model. In addition, FP-MobileNet weighs only occupy 13.7 MB with high recognition rate. It takes into account both accuracy and delay, and is suitable for promotion to mobile devices that require a lightweight model.

Dimensionality reduction plays an important role in big data analysis and visualization. Many dimensionality reduction techniques with probabilistic distribution models rely on the optimizaition of cost function between low-dimensional model distribution and high-dimensional real distribution. The key issue of this type of technology is to efficiently construct the probabilistic distribution model representing the feature of original high-dimensional dataset most. In this paper, Wasserstein distance was introduced to dimensionality reduction, and a novel method named Wasserstein Embedded Map (W-map) was presented for high-dimensional data reduction and visualization. W-map converts dimensionality reduction problem into optimal transportation problem by constructing the similar Wasserstein flow in the high-dimensional dataset and its corresponding low-dimensional representation, and then the best matched low-dimensional visualization was found by solving the optimal transportation problem of Wasserstein distance. Experimental results demonstrate that the presented method performs well in dimensionality reduction and visualization for high-dimensional data.

To solve high energy and time delay cost problems caused by wormhole detection in Ad Hoc networks, a light-weighted wormhole detection method, using less time delay and energy, was proposed. The method used the neighbor number of routing nodes to get a set of abnormal nodes and then detect the presence of a wormhole by using the neighbor information of abnormal node when routing process was completed. The simulation results show that the proposed method can detect wormhole with less number of routing query. Compared with the DeWorm (Detect Wormhole) method and the E2SIW (Energy Efficient Scheme Immune to Wormhole attacks) method, it effectively reduces the time delay cost and energy cost.

In view of the update problem of the trust value in Wireless Sensor Network (WSN), a trust model based on Fuzzy Prediction (FP), called RMFP, was proposed. The behavior of nodes was described by using fuzzy mathematics theory method, and the fuzzy membership degree was converted by the fuzzy membership functions. Finally, the trust value was achieved by integrating the fuzzy membership degrees. The simulation results show that the accuracy of trust value is increased by 10.8%, and the judgment speed of suspected nodes is increased by two times. This shows that the effect on accuracy and speed of discovering, eliminating malicious node is more significant, especially for the judgment of the pre-made malicious nodes of high trust value.

As a fast-growing contributor to Internet service, online videos make ISPs and video providers pay more attention to the Quality of user Experience (QoE). The existing routing algorithms could only guarantee that parameters of path such as delay and jitter, in accordance with the demands of Quality of Service (QoS) restriction. However, they are not able to reflect the QoE-related information directly. As a result, these kinds of algorithms cannot meet the demand of showing QoE. This paper proposed a QoE-optimal routing algorithm called QoE_DSP, based on QoE estimation model. By taking advantage of two properties of QoE parameter, decomposability and nondecreasing characteristics from the analysis on relationship between QoE and QoS, the authors designed QoE_DSP, which owned a polynomial time complexity of (V log V+E). According to the experiments and results analysis, this algorithm can guarantee the selected path meets the demand of QoE, while it also has a strong computational scalability.

In the cloud storage network, the security and integrity of data are the major concerns of clients. Taking full consideration of the security requirements for cloud storage network, a new Cloud Storage-Data Integrity Verification (CS-DIV) protocol was proposed. The clients uploaded files and tags to the servers and then did random check; the servers returned proofs and the clients judged the result. This protocol could not only ensure the integrity of data in the cloud storage effectively, but also resist the cheat from the un-trusted servers and the attack from the malicious clients, and then improved the reliability and stability of the whole cloud storage system. The simulation experimental results show that the proposed protocol realizes the protection of data integrity at low cost of storage, communication and delay.

Intensity inhomogeneity often exists in Magnetic Resonance (MR) images, which is due to the smooth bias field caused by the deficiency of the device. Traditional intensity-based segmentation algorithms often assume the uniform intensity belonging to the object and background, respectively. Therefore, these algorithms fail to successfully segment image with intensity inhomogeneity. This paper proposed a local region fitting model for simultaneous segmentation and bias correction. The model is built based on the intensity property in the local region to build an energy function with respect to the intensity, bias field function and the region indicating function. Then, this energy function was optimized with respect to the intensity, bias field and the indicating function, respectively. The segmentation and bias field estimation would be conducted simultaneously finally. The experimental results on the real MR brain images demonstrate the advantages of the proposed method over variational level set approach.

One of the critical problems in distributed network performance monitoring is that how to place proper number of monitoring agents in proper locations in a managed network can improve monitoring efficiency,that is,decrease monitoring traffic and response time.Our algorithm was triggered when network topology changed,or nodes and links failures occurred.By adjusting some agents’ deployment,it could draw the whole monitoring system back to proximate optimization state.Simulation shows that this algorithm can add flexibility and scalability to the proposed monitoring system.

After analyzing the behavior of accessed users, browsers and existing Markov prediction models, a TOP-N selective Markov model was presented to predict user next requests. The TOP-N consisted of URLs which requested time was over N in Web logs. The Markov chains were made up of user visit sessions. If the session which user visited currently matched one of the Markov chains, the next URL of this Markov chains in TOP-N would be prefetched in local cache. The experiment results show that this model can achieve dramatic improvement on predictive accuracy and get a good hit ratio with reducing the traffic load to some extent.