Search

Dictionary learning is widely utilized in pattern recognition, and analysis dictionary learning is a prevalent image classification method. However, its classification performance still has much room for improvement. Constructing the powerful discriminative constraint by exploiting the intrinsic characteristics of sample is an effective approach for enhancing the performance, thus how to design an effective constraint is a problem worth studying. On the other hand, some analysis dictionary learning models incorporate the classification error constraint. However, this constraint always adopts a strict binary matrix as the target matrix, which is harmful to the improvement of classification performance. To solve these issues, we propose an interactive and discriminative analysis dictionary learning for image classification, The ordinal locality preserving technique is utilized to to preserve the topology information of the samples, and Fisher constraint is applied to promote the discrepancy of inter-class coding coefficients and the similarity of intra-class coding coefficients. Furthermore, the target matrix is relaxed by exploiting ℓ 21 norm, which can provide more freedom to the classifier parameter. Finally, comparative experiments on different types of data sets show the efficacy of the proposed method, For 15 Scene dataset, the proposed IDADL improve the classification accuracies by 1.5%, 0.9% and 0.6% over FDDL, SLCADL and RADPL, respectively. Besides, the performance of the proposed model presents superiority over some classical deep models, For CMU PIE, the proposed IDADL promote the classification accuracies by 4.3% and 1.6% over AlexNet and VGG, respectively. [ABSTRACT FROM AUTHOR]

Fully capturing valid complementary information in multi-view data enhances the connection between similar data points and weakens the correlation between different data point categories. In this paper, we propose a new multi-view clustering via dual-norm and Hilbert-Schmidt independence criterion (HSIC) induction (MCDHSIC) approach, which can enhance the complementarity, reduce the redundancy between multi-view representations, and improve the accuracy of the clustering results. This model uses the HSIC as the diversity regularization term to capture the nonlinear relationship between different views. In addition, l1-norm and Frobenius norm constraints are imposed to obtain a subspace representation matrix with inter-class sparsity and intra-class consistency. Moreover, we also designed a valuable approach to optimizing the proposed model and theoretically analyzing the convergence of the MCDHSIC method. The results of extensive experiments conducted on five challenging data sets show that the proposed method objectively achieves a highly competent performance compared with several other state-of-the-art multi-view clustering methods. [ABSTRACT FROM AUTHOR]

In order to explore the importance of the hypergraph regularization and the Tikhonov regularization in multi-view clustering, this paper proposes a novel multi-view clustering model, termed as low-rank tensor multi-view subspace clustering via collaborative regularization (LT-MSCCR). The LT-MSCCR model introduces the idea of tensor. The tensor is designed to represent the result of superimposing the subspace representation matrix, which is used to explore the high order correlations between different views, and the low-rank restriction is performed for this tensor to effectively reduce redundant information. Furthermore, we adopt the hypergraph regularization to mine effective geometric information in multi-view data. Meanwhile, we also impose the Tikhonov regularization constraint on the subspace representation matrix so as to improve the smoothness of the subspace representation matrix and enhance the recognition performance of the proposed method. In addition, we also designed a valuable approach to optimizing the proposed model and theoretically analyzing the convergence of the LT-MSCCR method. The experimental results on some datasets show that the proposed model is better than many advanced multi-view clustering methods. [ABSTRACT FROM AUTHOR]

In order to mine the manifold information in different views, a hypergraph regularized low-rank tensor multi-view subspace clustering via L1 norm constraint (HLRT-MSCL1) is proposed. The tensor in this method is composed of subspace representation matrix of different views, which is used to mine the high order correlations between various views, and the low-rank restriction is performed for the tensor to effectively decrease the redundant data of the subspace representation. In addition, the HLRT-MSCL1 method considers the local geometric information of different views and imposes sparse constraint on the subspace representation matrix, so that its recognition performance is improved to a certain extent. Moreover, an effective method for solving the HLRT-MSCL1 model is also designed, and its convergence is discussed. The results of extensive experiments conducted on five challenging data sets show that the proposed method objectively achieves a highly competent performance compared with several other state-of-the-art multi-view clustering methods. [ABSTRACT FROM AUTHOR]

Via micro- and nano-structure design of a coating layer on the bottom of ceramic pots with copper powders and low melting point glass powders, a simple and low-cost solution for fabricating high-efficiency induction heating ceramic pots without changing the induction cookers has been provided. The influences of sintering temperature and glass content on the heating efficiency of the copper layer were studied. The hysteresis loop, resistivity, and microstructure of the copper layer were investigated using a vibrating sample magnetometer, a field-emission scanning electron microscope, and a high-resolution transmission electron microscope, respectively. Microstructure observation revealed that an insulating glass film with 1.06 nm width was formed between the adjacent copper particles in the sintered copper layer. The experimental results showed that the sintered layer made with copper powder slurry had a resistivity of 22.67 × 10–8 Ωm, which is 12.95 times that of a pure copper plate. The saturation magnetization of the sintered layer under the static magnetic field was about 3.55 × 10–2 emu/g, which is 3.38 times that of a copper plate. A ceramic pot with a sintered copper layer takes about 510 s to heat 500 mL of water from room temperature to 100°C under the power of 500 W. Its heating efficiency is comparable to that of a ferromagnetic stainless-steel pot. [ABSTRACT FROM AUTHOR]

In this paper, to extract the manifold information from multi-view data and enhance the clustering performance of a multi-view learning method, the multi-view subspace enhanced representation of manifold regularization and low-rank tensor constraint (MSERMLRT) method is introduced. Our model uses a tensor to explore the correlation between views. The tensor is constrained with a low-rank, and the purpose of such processing is to reduce the redundant information of the learned subspace representation. This model also uses the manifold information from multi-view data and imposes a sparse constraint on the product of itself and the transpose of the subspace representation matrix to enhance the diagonal block structure of the subspace representation, thereby improving its clustering effect to a certain extent. We also designed a helpful method for solving the MSERMLRT model and analyzed the convergence of our approach both theoretically and experimentally. The clustering performance on certain challenging datasets indicate that the MSERMLRT model is superior to many other advanced multi-view clustering methods. [ABSTRACT FROM AUTHOR]

The non-negative matrix factorization (NMF) method is applied in many fields, including pattern recognition, visual analysis, and biomedicine. However, existing NMF methods acquire low dimensional representation directly from high dimensional data, and the low-rank part of the data contains the most useful information. Furthermore, the NMF method is vulnerable to noise interference and is not adequately robust. Therefore, a robust semi non-negative low-rank graph embedding algorithm via the L21 norm (RSNLGEL21) is proposed to address the stated problems. The algorithm considers the effective low-rank structure and geometric information of the original data and constrains the coefficient matrix to be non-negative; however, it does not impose a non-negative constraint on the base matrix. In addition, the L21 norm is introduced into the graph embedding and data reconstruction functions slightly to improve its recognition ability and robustness. Moreover, this paper also provides an iterative formula to the RSNLGEL21 algorithm and the corresponding proof of convergence and to analyze its computational complexity. Experiments on the ORL, FEI, AR, and FERET image datasets shows that the RSNLGEL21 algorithm has certain advantages in clustering performance. [ABSTRACT FROM AUTHOR]

Catalytic reduction of NO by material surface attracts growing interests due to its promising applications in air purification. Here, the adsorption and dissociation behaviors of single NO molecule on the Be(0001) surface were systematically investigated using first-principles calculations. Through potential energy surface calculations, we found that NO molecule can adsorb on the surface without any energy barrier along the vertical channels with N end-on orientation, among which adsorption on the hcp hollow site of the Be(0001) surface is the most energetically favorable. Moreover, the dissociation of NO molecule along the fcc hollow channel was investigated, and it was found that the dissociation process of NO is an activated type with an energy barrier of 0.34 eV, indicating that the reduction of NO on the Be(0001) surface is energetically probable. The electronic structure analyzing of the most stable chemisorption state that locates on the hcp site, along with the sizeable charge redistribution, reveals that the chemisorption of NO on the Be(0001) surface is dominated by the interaction between the molecular orbital 2 π ∗ of NO molecule and electronic state p of surface Be atom. [ABSTRACT FROM AUTHOR]