Focusing on the challenge task for mining complementary information in different levels of features in the deep subspace clustering problem, based on the deep autoencoder, by exploring complementary information between the low-level and high-level features obtained by the encoder, a Diversity Represented Deep Subspace Clustering (DRDSC) algorithm was proposed. Firstly, based on Hilbert-Schmidt Independence Criterion (HSIC), a diversity representation measurement model was established for different levels of features. Secondly, a feature diversity representation module was introduced into the deep autoencoder network structure, which explored image features beneficial to enhance the clustering effect. Furthermore, the form of loss function was updated to effectively fuse the underlying subspaces of multi-level representation. Finally, several experiments were conducted on commonly used clustering datasets. Experimental results show that on the datasets Extended Yale B, ORL, COIL20 and Umist, the clustering error rates of DRDSC reach 1.23%, 10.50%, 1.74% and 17.71%, respectively, which are reduced by 10.41, 16.75, 13.12 and 12.92 percentage points, respectively compared with those of Efficient Dense Subspace Clustering (EDSC), and are reduced by 1.44, 3.50, 3.68 and 9.17 percentage points, respectively compared with Deep Subspace Clustering (DSC), which indicates that the proposed DRDSC algorithm has better clustering effect.
In order to accelerate rendering of 3D scenes by path tracing, a visual?saliency?driven reuse algorithm of indirect lighting in 3D scene rendering was proposed. Firstly, according to the characteristics of visual perception that the regions of interest have high saliency, while other regions have low saliency, a 2D saliency map of the scene image was obtained, which consists of color information, edge information, depth information and motion information of the image. Then, the indirect lighting in the high?saliency area was re?rendered, while the indirect lighting of the previous frame was reused in the low?saliency area under certain conditions, thereby accelerating the rendering. Experimental results show that the global lighting effect of the image generated by this method is real, and the rendering speed of the method is improved in several experimental scenes, and the speed can reach up to 5.89 times of that of the high?quality rendering.
Aiming at the problem of slowness of the current meta-heuristic algorithms when solving Travelling Salesman Problem (TSP) in combinatorial optimization problems, a multi-scale adaptive quantum free particle optimization algorithm was proposed based on the inspiration of the wave function in quantum theory. Firstly, the particles representing the city sequences were randomly initialized in the feasible region as the initial search centers. Then, the new solution was obtained by taking each particle as the center to perform the sampling with uniformly distributed function and exchanging the city numbers in the sampling positions. Finally, according to the comparison result of the new solution with the optimal solution in the previous iteration, the search scale was adaptively adjusted, and the iterative search was carried out at different scales until the end condition of the algorithm was satisfied.The algorithm was compared with Hybrid Particle Swarm Optimization (HPSO) algorithm, Simulated Annealing (SA), Genetic Algorithm (GA) and Ant Colony Optimization(ACO) algorithm on TSP. The experimental results show that the multi-scale quantum free particle optimization algorithm is suitable for solving combinatorial optimization problems, and increases the solving speed by over 50% on average compared with the current better algorithms on the TSP datasets.
Since most malwares are designed using decentralized architecture to resist detection and countering, in order to fast and accurately detect Peer-to-Peer (P2P) bots at the stealthy stage and minimize their destructiveness, a real-time detection system for stealthy P2P bots based on statistical features was proposed. Firstly, all the P2P hosts inside a monitored network were detected using means of machine learning algorithm based on three P2P statistical features. Secondly, P2P bots were discriminated based on two P2P bots statistical features. The experimental results show that the proposed system is able to detect stealthy P2P bots with an accuracy of 99.7% and a false alarm rate below 0.3% within 5 minutes. Compared to the existing detection methods, this system requires less statistical characteristics and smaller time window, and has the ability of real-time detection.
In we media platform such as microblog, emergency has such characteristics as suddenness and having multiple bursting points. Thus, it brings difficulty to emergency detection. Thus, this paper proposed a method of bursty events detection based on sentiment filter. Firstly, the topic was mapped as a hierarchical model according to the method. Then, dynamic adjustment of the model characteristics was made in a timing-driven way so as to detect the new topics of the information. Based on it, the method analyzed the user's emotional attitude toward such topics. The topics were divided into positive and negative emotion tendencies according to the user's emotional attitude. Additionally, the topic full of negative emotion tendency was regarded as emergent topic. The experimental results show that the accuracy and recall of the proposed method are all increased about 10% compared with baseline.
For the difficulty of expressing spatial context in classification of high resolution remote sensing imagery, a new multi-scale Conditional Random Field (CRF)model was proposed here. Specifically, a given image was represented as three superpixel layers respectively being region, object and scene from fine to coarse firstly. Then features were extracted layer-by-layer, and those features from the three layers were associated with each other to form a feature vector for each node in region layer. Secondly, Support Vector Machine (SVM) was adopted to define association potential function, and Potts model weighted by feature contrast function was used to define interaction potential function of CRF model, thus a layer-by-layer feature associative and multi-scale SVM-CRF model was formed. To confirm the effectiveness of the proposed model in classification, experiments on two complex scenes from Quickbird remote sensing imagery were developed. The results show that the proposed model achieves an improved accuracy averagely 2.68%, 2.37%, 3.75% higher than that of SVM-CRF model based on either region, object or scene layer, also it consumes less time in classification.
It is difficult for the existing methods to get overall sentiment orientation of the comment text. To solve this problem, the method of multi-document sentiment summarization based on Latent Dirichlet Allocation (LDA) model was proposed. In this method, all the subjective sentences were extracted by sentiment analysis and described by LDA model, then a summary was generated based on the weight of sentences which combined the importance of words and the characteristics of sentences. The experimental results show that this method can effectively identify key sentiment sentences, and achieve good results in precision, recall and F-measure.