Aiming at the problem that the influence of time and space information of the location was not considered when making the location recommendation by Gated Recurrent Unit (GRU) of recurrent neural network, the spatiotemporal-aware GRU model was proposed. In addition, aiming at the noise problem generated by the unrelated check-in data in check-in sequence, the next location recommendation method of SpatioTemporal-aware GRU and Attention (ST-GRU+Attention) was proposed. Firstly, time gate and distance gate were added in the GRU model by counting the time slot and distance gap between two locations. The influence of time and space information on recommending next location was controlled by setting the weight matrices. Secondly, the attention mechanism was introduced. The attention weight coefficients of the user were obtained by calculating the attention weight scores of the user preferences, and the personalized preference of the user was obtained. Finally, the objective function was constructed and the model parameters were learned by Bayesian Personalized Ranking (BPR) algorithm. The experimental results show that the accuracy of ST-GRU+Attention is improved significantly compared to the recommendation methods of Factorizing Personalized Markov Chain and Localized Region (FPMC-LR), Personalized Ranking Metric Embedding (PRME) and Spatial Temporal Recurrent Neural Network (ST-RNN), and the precision and recall of ST-GRU+Attention are increased by 15.4% and 17.1% respectively compared to those of ST-RNN which is the best of the three methods. The recommendation method of ST-GRU+Attention can effectively improve the effect of next location recommendation.

To solve the problem that Delegated Proof of Stake (DPoS) consensus mechanism has malicious nodes not eliminated in time due to inactive voting and long voting cycle, an improved scheme of DPoS consensus mechanism based on fusing mechanism, credit mechanism and standby witness node was proposed. Firstly, fusing mechanism was introduced to provide the function of negative vote to quicken kicking out evil nodes. Secondly, credit mechanism was introduced to set credit scores and credit grades for nodes, the credit scores and grades of nodes were dynamically adjusted by monitoring the behavior of nodes, therefore the difficulty of obtaining votes for evil nodes was increased. Finally, standby witness node list was added to fill in the vacancy in time after witness right of evil node being cancelled. A test blockchain system based on the improved scheme was built, and the availability and effectiveness of the improved scheme were verified by experiments. The experimental results show that the blockchain based on the improved DPoS consensus mechanism can eliminate the evil nodes in time and is suitable for most scenarios.

To solve the signal demodulation problem of asymmetric Paired Carry Multiple Access (PCMA) composed of the same frequency of main station and small station signals, a framework to realize this kind of signal demodulation was constructed. Parameter estimation is an indispensable part in the realization of two-way signal separation and demodulation for asymmetric PCMA communication systems. For the estimation accuracy of amplitude parameters, a searching amplitude estimation algorithm based on fourth-power method was proposed. Firstly, the demodulation model for asymmetric PCMA systems was established and the basic assumptions were made. Then the phase errors under different assumptions were compared with each other and the influence of phase error on the amplitude estimation algorithm was analyzed. Finally, a new amplitude estimation algorithm was proposed. Experimental results show that, under same Signal-to-Noise Ratio (SNR), the demodulation performance of the small station signal under normal phase error is inferior to its demodulation performance under mean value condition. When the order of magnitude of the Bit Error Rate (BER) is 10 ^-4, the demodulation performance of small station signal is improved by 1 dB with the improved algorithm, proving that the improved algorithm is better than fourth-power method.

An algorithm combining Structured Occlusion Coding (SOC) with Extreme Learning Machine (ELM) was proposed to deal with the occlusion problem in face recognition. Firstly, the SOC was used to remove the occlusion from the image and separate the oclusion from the human face. At the same time, the position of the occlusion was estimated by the Local Constraint Dictionary (LCD), and an occlusion dictionary and a face dictionary were established. Then, the established face dictionary matrix was normalized, and the ELM was used to classify and identify the normalized data. Finally, the simulation results on the AR face database show that the proposed method has higher recognition rate and stronger robustness for different types of occlusions and images with different regions occluded.

Aiming at the over-segmentation problem caused by inconsistency of large-format, high-resolution and inconsistency of parcel size in extraction of Unmanned Aerial Vehicle (UAV) remote sensing image of farmland scene, an automatic extraction process for land boundary based on multi-scale segmentation was proposed. In this process, the block segmentation strategy was adopted under the framework of Multi-scale Combinatorial Grouping (MCG) segmentation method. The optimal ground sampling distance was selected by comparing experimental research and optimal segmentation scale was selected by analyzing the variation curve of boundary extraction accuracy with scale, therefore automatic extraction process of parcel boundaries was achieved. Experiments were conducted on the data collected from Xiantao City, Hubei Province. The experimental results show that the most suitable ground sampling distance for extracting land parcel boundary is about 30 cm and the optimal segmentation scale is[0.2,0.4]. The accuracy of land parcel boundary extraction can be more than 90%. In addition, the proposed method can accurately extract large-scale agricultural parcel boundary and also can provide a reference for later aerial program of agriculture UAV.

Aiming at the problem of poor detail and visibility in current color coding methods of single polarization Synthetic Aperture Radar (SAR), a color feature coding method was proposed. Firstly, texture features were extracted from a single-polarized SAR image. Secondly, each feature was quantized to 0 to 255. Then an RGB color was assigned to each gray level to generate a color feature map. Finally, the importance of features calculated by random forest was sorted; the pseudo-color graphs were generated by each three dimensional feature corresponding to the R, G, and B channels. Based on the presented color feature coding method, a new classification method was proposed. Firstly, the pseudo color map with the best geographical separability was selected according to the visual effect, and then segmented by the Statistical Region Merging (SRM) segmentation algorithm. Secondly, all the RGB pseudo color maps were used as the classification features, and a random forest was used as the classifier and obtain the preliminary results. At the end, a relative majority vote was made on the preliminary results and the final classification results were obtained. In the method verification, two sets of TerraSAR-X single-polarization SAR data were used. By comparing the corresponding grayscale image with HIS-based color coding method, the color image information entropy generated by the proposed color feature coding method was greatly improved, and the classification accuracy of each type of ground features for two data sets was greatly improved. It is demonstrated that the proposed algorithm preserves more details for more color information, and it is more conducive to visualization and terrain classification, which indicating the proposed color feature coding method is feasible.

In recent years, test question resources in online education has grown at an explosive rate. It is difficult for students to find appropriate questions from the mass of question resources. Many test question recommendation methods for students have been proposed to solve this problem. However, many problems exist in traditional test question recommendation methods based on unified probalilistic matrix factorization; especially information of student knowledge points is not considered, resulting in low accuracy of recommendation results. Therefore, a kind of personalized test question recommendation method based on unified probalilistic matrix factorization was proposed. Firstly, through a cognitive diagnosis model, the student knowledge point mastery information was obtained. Secondly, the process of unified probalilistic matrix factorization was executed by combining the information of students, test questions and knowledge points. Finally, according to the difficulty range, the test questions were recommended. The experimental results show that the proposed method gets the best recommedation results in the aspect of accuracy of question recommendation for different range of difficulty, compared to other traditional recommendation methods, and has a good application prospect.

In order to further improve the execution efficiency of Frequent Pattern-Growth (FP-Growth) algorithm on Spark platform, a new parallel FP-Growth algorithm based on Spark, named BFPG (Better Frequent Pattern-Growth), was presented. Firstly, the grouping strategy F-List was improved in the size of the Frequent Pattern-Tree (FP-Tree) and the amount of partition calculation to ensure that the load sum of each partition was approximately equal. Secondly, the data set partitioning strategy was optimized by creating a list P-List, and then the time complexity was reduced by reducing the traversal times. The experimental results show that the BFPG algorithm improves the mining efficiency of the parallel FP-Growth algorithm, and the algorithm has good scalability.

Focusing on the problems of small samples, large face pose changes, occlusion and complex background, under unconstrained condition, a face verification method based on 3D frontalization and similarity learning was proposed. Firstly, the 3D frontalization progress was applied to generate the frontal face of the face image. Secondly, the complex background was removed by cropping the relevant face regions. Finally, a similarity learning method based on intra-personal subspace was applied to measure the similarity of the image pairs. Experiments were conducted on several databases that were built up by preprocessing the Labeled Faces in the Wild (LFW) database. the difference between these databases and original LFW is their images have been preprocessed. In the experiment with Local Ternary Pattern (LTP) descriptor as the feature extraction method and 625 training image pairs, the recognition rate of the proposed algorithm Similarity Learning over subspace (sub-SL) was 15.6% and 8.4% higher than that of Metric Learning over subspace (sub-ML) and Similarity Metric Learning over subspace (sub-SML) respectively. Experimental results show that the proposed algorithm can effectively improve the accuracy of face verification under unconstrained conditions.

Separability Index (SI) can be used to select effective classification features, but in the case of multi-dimensional features and good separability of geology, the use of separability index for feature selection can not effectively remove redundancy. Based on this, a method of feature selection and multi-layer Support Vector Machine (SVM) classification was proposed by using separability index and Sequential Backward Selection (SBS) algorithm. Firstly, the classification object and features were determined according to the SIs of all the ground objects under all the features, and then based on the classification accuracies of the objects, the SBS algorithm was used to select the features again. Secondly, the features of next ground objects were determined by the separability index of remaining objects and the SBS algorithm in turn. Finally, the multi-layer SVM was used for classification. The experimental results show that the classification accuracy of the proposed method is improved by 2% compared with the method of multi-layer SVM classification where features are selected only based on the SI, and the classification accuracy of all kinds of objects is higher than 86%, and the running time is half of the original method.

In order to solve the optimization problem of dynamic task allocation in multi-robot system, a new quadratic task allocation algorithm based on Pareto improvement was proposed based on the initial task allocation of contract net. When the fire fighting task was performed by the multi-robot system in parallel, firstly, the multiple robots were divided into several sub-group through the initialization of task allocation. Then, a fire fighting task was undertook by a subgroup, and the robots needed to be migrated were determined by the Pareto improvement of the subgroup and its nearest subgroup while the subgroup performing the task to achieve the Pareto optimality between the two subgroups. Finally, the global Pareto optimality was achieved by the Pareto improvement of all subgroups with posterior binary tree traversal. The theoretical analysis and simulation results show that, compared with reinforcement learning algorithm and ant colony algorithm, the fire fighting time of the proposed algorithm is reduced by 26.18% and 37.04% respectively. And compared with the traditional contract net method, the proposed algorithm can perform the fire fighting task efficiently in time, and also has the obvious advantage in system revenue.

In order to make full use of the ability of of Synthetic Aperture Radar (SAR) images with different polarization features for characterizing different types of ground objects, an analysis and classification approach of polarimetric SAR feature based on multi-layer Support Vector Machine (SVM) classifier was proposed. Firstly, the optimal feature subsets suitable for different terrain types were determined through the feature analysis. Then, the method of hierarchical classification tree was used for SVM classification step by step according to the feature subset of each object type.Finally, the overall final result was obtained. The experimental results of RadarSAT-2 polarimetric SAR image classification show that, the classification accuracy of the proposed approach is approximately 85% for four kinds of ground objects such as water area, cultivated land, forest land and urban area and the overall classification accuracy is up to 86%. The proposed approach can make full use of the characteristics of the different ground object target types, improve the classification accuracy and reduce the time complexity.

The parameter of traditional regularized Super Resolution (SR) reconstruction model is difficult to choose:the higher parameter value results in blurred reconstruction and the fading of edge and detail, while the lower parameter value weakens the denosing ability. A double regularization parameters super resolution reconstruction method based on structure tensor was proposed. Firstly, smooth region and edge was detected by using local structure tensor. Secondly, the Total Variation (TV) was weighted with the priori information of difference curvature. Finally, two different parameters toward smooth region and edge were used to reconstruct super resolution image. The experimental data show that the proposed algorithm can improve the Peak Signal-to-Noise Ratio (PSNR) of 0.033-0.11 dB, and get better reconstruction results. The proposed algorithm can effectively improve the reconstruction effect of Low Resolution (LR) video frames, and can be applied to LR video enhancement, license plate recognition and the interest target enhancement in video surveillance, etc.

Current neighborhood rough sets have been usually used to solve complete information system, not incomplete system. In order to solve this problem, an extended tolerance relation was proposed to deal with the incomplete mixed information system, and associative definitions were provided. The degree of complete tolerance and neighborhood threshold were used as the constraint conditions to find the extended tolerance neighborhood. The attribute importance of the system was got by the decision positive region within the neiborhood, and the attribute reduction algorithm based on the extended tolerance relation was proposed, which was given by the importance as the heuristic factor. Seven different types of data sets on UCI database was used for simulation, and the proposed method was compared with Extension Neighborhood relation (EN), Tolerance Neighborhood Entropy (TRE) and Neighborhood Rough set (NR) respectively. The experimental results show that, the proposed algorithm can ensure accuracy of classification, select less attributes by reduction. Finally, the influence of neighborhood threshold in extended tolerance relation on classification accuracy was discussed.

Since the algorithm of attribute reduction based on positive region is based on the thought of lower approximation, it just considers the right distinguished samples. Using the thought of upper approximation and the concept of neighborhood information granule, the distinguished object set with its basic characteristics was designed and analyzed, then the new attribute importance measurement based on distinguished object set and heuristic attribute reduction algorithm was proposed. The proposed algorithm considered both the relative positive region of information decision table and the influence on boundary samples when growing condition attributes. The feasibility of the algorithm was discussed by instance analysis, and the comparative experiments on UCI data set with attribute reduction algorithm based on positive region were carried out. The experimental results show that the proposed attribute reduction algorithm can get better reduction, and the classification precision of sample set can remain the same or has certain improvement.

Focusing on the underdeveloped robustness when the existing extended rough set model encounters the noise for the incomplete information system, the necessity of adjusting the size of basic knowledge granule as well as introducing the relative degree of misclassification was analyzed. Then the Variable Precision Rough Set model based on Variable-Precision Tolerance Relation (VPRS-VPTR) was established on the basis of the object connection weight matrix, which was proposed according to the lack probability of system attribute value. Moreover, the properties of the VPRS-VPTR model were discussed, the classification accuracy under the basic knowledge granule size and the relative degree of misclassification was analyzed, the corresponding algorithm was depicted and the time complexity analysis was given afterwards. The experimental results show that the VPRS-VPTR model has higher classification accuracy compared with some other research about the expanded rough set, and the change trend of the classification accuracy is similar for the train set and the test set of several groups of incomplete data sets in UCI database. It proves that the proposed model is more precise and flexible, and the algorithm is feasible and effective.

Aiming at the problem that the security surveillance cameras have been hidden by leaves, a leaf occlusion detection algorithm based on Support Vector Machine (SVM) was proposed. The algorithm contains three steps. First, the regions of the leaf existing in the video were segmented. The accumulated frame subtraction method was applied to achieve this purpose. Second, the color and area information of the whole video image and the segmented regions were extracted as the key features. Third, these features were used for modeling and detecting obstacle occlusion by SVM. For all the collected samples, the detection accuracy of this method can reach up to 84%. The experimental results show that the proposed algorithm can detect the leaf occlusion in security surveillance video effectively.

For the fault-tolerance shortage of CFD (Computational Fluid Dynamics)-oriented application development framework, a new fault-tolerance period optimization method was proposed. The method computed the ideal best fault-tolerance period based on the probability model of system's faults, and online determined the occasion of real check points with the consideration of CFD fields output characteristic. The experimental results of three applications show that on the systems with different mean time between faults, compared with the fault-tolerance method based on performing fault-tolerance between fixed steps, the proposed method can always get the best fault-tolerance overheads. Based on this method, user can set the fault-tolerance period with framework interfaces conveniently and reduce the fault-tolerance overheads.

The seizure detection is important for the localization and classification of epileptic seizures. In order to solve the problem brought by large amount of data and high feature space in EEG (Electroencephalograph) for quickly and accurately detecting the seizures, a method based on max-Relevance and Min-Redundancy (mRMR) criteria and Extreme Learning Machine (ELM) was proposed. The time-frequency measures by Short-Time Fourier Transform (STFT) were extracted as features, and the large set of features were selected based on max-relevance and min-redundancy criteria. The states were classified using the extreme learning machine, Support Vector Machine (SVM) and Back Propagation (BP) algorithm. The result shows that the performance of ELM is better than SVM and BP algorithms in terms of computation time and classification accuracy. The classification accuracy rate of interictal durations and seizures can reach more than 98%, and the computation efficiency is only 0.8s. This approach can detect epileptic seizures accurately in real-time.

Non-interactive deniable authentication protocol can enable the receiver to identify the source of a received message and prevent a third party from identifying the source of the message, which is very suitable to be used in E-commerce and E-government. Based on computational Diffie-Hellman assumption, using bilinear pairing, a new identity-based deniable authentication protocol was constructed. The security of the scheme was proved under the random oracle model. The analytical results show that the new proposed protocol can resist the forgery attack, spoofing attack, middle attack and replay attack. This protocol is identity-based, which means it needs no certificate and has a simple key management. On the other hand, it is efficient in communications and computation, and its implementation is simple, so that it could be implemented in mobile devices with low power and small processor.

Using bilinear pairing, a new identity-based signcryption scheme was proposed in this paper. Under the assumption that the Computational Diffie-Hellman (CDH) problem is hard, the newly proposed scheme had been proved to be secure against the existing unforgeability on adaptively chosen message/ciphertext and identity attack in random oracle model. The advantage of the proposed scheme is that it is identity-based which needs no certificates so that it has a simple key management. In addition, the proposed scheme can provide public verifiability, and it allows a third party to convince that the signcryption is valid for the given message without providing the receiver's private key.

Liu's online/offline signcryption scheme (LIU J K, BAEK J, ZHOU J. Online/offline identity-based signcryption re-visited. Inscrypt'10: Proceedings of the 6th International Conference on Information Security and Cryptology. Berlin: Springer-Verlag, 2010:90-102) was cryptanalyzed and improved in this paper. By studying the unforgeability of Liu's scheme, this paper proved that Liu's scheme would be not unforgeable against adaptive chosen message and identity attack. To overcome the security problem in Liu's scheme, a modified scheme was proposed, and the modified scheme was proved to be a secure Identity-based online/offline signcryption scheme.

Large-scale computing and high mis-classification rate are two disadvantages of Loopy Belief Propagation (LBP) algorithm for image segmentation. A fast image segmentation method based on LBP algorithm was proposed. At first, a local region Gibbs energy model was built up. Then the region messages were propagated by LBP algorithm. In order to improve the running speed for LBP algorithm, an efficient speedup technique was used. At last, the segmentation result was obtained by the Maximum A Posterior (MAP) criterion of local region Gibbs energy. The experimental results show that the proposed algorithm not only obtains more accurate segmentation results, especially to noise or texture image, but also implements more fast.