A predator–prey system with stage structure and time delay for the prey is investigated. By analyzing the corresponding characteristic equations, the local stability of a positive equilibrium and two boundary equilibria of the system is discussed, respectively. By using persistence theory on infinite dimensional systems and comparison argument, respectively, sufficient conditions are obtained for the global stability of the positive equilibrium and one of the boundary equilibria of the proposed system. Further, the existence of a Hopf bifurcation at the positive equilibrium is studied. Numerical simulations are carried out to illustrate the main results.

This paper proposes using a previously well-trained deep neural network (DNN) to enhance the i-vector representation used for speaker diarization. In effect, we replace the Gaussian mixture model typically used to train a universal background model (UBM), with a DNN that has been trained using a different large-scale dataset. To train the T-matrix, we use a supervised UBM obtained from the DNN using filterbank input features to calculate the posterior information and then MFCC features to train the UBM instead of a traditional unsupervised UBM derived from single features. Next we jointly use DNN and MFCC features to calculate the zeroth- and first-order Baum–Welch statistics for training an extractor from which we obtain the i-vector. The system will be shown to achieve a significant improvement on the NIST 2008 speaker recognition evaluation telephone data task compared to state-of-the-art approaches.

Built by the royal family in the Ming Dynasty, the Da Bao En Temple is known as the greatest temple in ancient Nanjing, whose predecessor is the Chang Gan Temple built in the Six Dynasties. Archaeological excavations of the Da Bao En Temple and the underground palace of Chang Gan Temple built in the North Song Dynasty (AD960-AD1127) have been appraised as one of the 10 greatest archaeological discoveries in 2010 in China. Many artefacts discovered in the underground palace have shown their important historical meanings and scientific values, such as the Qibao King Asoka Tower, golden and silver coffins, Buddhist relics, the silk textiles, glasses, and spices etc. In this paper, stereomicroscope, SEM-EDS and LA-ICP-AES are used to investigate chemical composition, microstructure and current preservation status for the unearthed glasswares. The results indicate that the glass bottle coded as TH1 and the glass calyx coded as TN5 are made of lead-silicate glass, while the chemical composition of the glass bottle coded as TN9 is quite distinct from that of native glasswares. All three articles have been weathered in some degrees. Given the shape, it is deduced that TH1 is a typical glassware used in burying Buddhist relics at that time, TN5 a domestic glassware with typical Sassanian style, and TN9 an imported Islamic glass, providing important information about culture exchanges between China and the foreign countries in the North Song Dynasty.

Oily water treatment has attracted the attention of many researchers. The development of effective and cheap oil/water separation materials is urgent for treating this problem. Herein, inspired by superhydrophobic typical plant leaves such as lotus, red rose and marigold, superhydrophobic and superoleophilic copper mesh was fabricated by etching and then surface modification with 1-dodecanethiol (HS(CH₂)₁₁CH₃). A rough silver layer is formed on the mesh surface after immersion. The obtained mesh surface exhibits superhydrophobicity and superoleophilicity and the static water contact angle was 153° ± 3°. In addition, the as-prepared copper mesh shows self-cleaning character with water and chemical stability. The as-prepared copper foam can easily remove the organic solvents either on water or underwater. We demonstrate that by using the as-prepared mesh, oils can be absorbed and separated, and that high separation efficiencies of larger than 92% are retained for various oils. Thus, such superhydrophobic and superoleophilic copper mesh is a very promising material for the application of oil spill cleanup and industrial oily wastewater treatment.

In this paper, we propose a fast and efficient learning approach called WELM based on Extreme Learning Machine and 3-D Wavelet Dynamic Co-Movement Analysis to enhance the speed and precision of big data prediction. 3-D Wavelet Dynamic Co-Movement Analysis is firstly employed to transform optimization problems from an original higher-dimensional space to a new lower-dimensional space while preserving the optimum of the original function, and then ELM is utilized to train and forecast the whole process. WELM model is used in the volatility of time series prediction. The forecasts obtained by WELM has been compared with ELM, PCA-ELM, ICA-ELM, KPCA-ELM, SVM and GARCH type models in terms of closeness to the realized volatility. The computational results demonstrate that the WELM provides better time series forecasts and it shows the excellent performance in the accuracy and efficiency.

There is an increasing consensus that software-defined networking may become a successful case to provide fine scalability and availability for industrial Internet, and it also brings new opportunities for the development of industrial cyber security. Aligning with the defense in depth strategy, this paper proposes a software-defined data flow detection and control approach for industrial Modbus/TCP communication. Furthermore, this approach designs a novel security strategy configuration service in SDN controllers to publish the flow control rules, and SDN switches match Modbus/TCP data flows with these flow control rules to detect and control abnormal communication behaviors. Specifically, a flow control rule database which stores all flow control rules of the entire control system is managed by SDN controllers, and a security flow table is maintained by each SDN switch according to different requirements of industrial communication. By using the DPI (Deep Packet Inspection) technology, this approach can run a deep analysis of Modbus/TCP packets according to the protocol specification, and block the improper control commands or undesired technology parameters. The qualitative analysis shows that the proposed approach possesses certain advantages and feasibilities.

In this paper, an innovative method called extreme learning machine with hybrid local receptive fields (ELM-HLRF) is presented for image classification. In this method, filters generated by Gabor functions and the randomly generated convolution filters are incorporated into the convolution filter kernels of local receptive fields based extreme learning machine (ELM-LRF). Extreme learning machine (ELM) is derived from single hidden layer feed-forward neural networks, and the parameters of its hidden layer can be generated randomly. As locally connected ELM, ELM-LRF directly processes information with strong correlations such as images and speech. In this paper, two main contributions are proposed to improve the classification performance of ELM-LRF. First, the Gabor functions are used as one kind of convolution filter kernels of ELM-HLRF to execute image classification. Second, we use a data augmentation method to preprocess training images to avoid overfitting. Experiments on the Outex texture dataset, the Yale face dataset, the ORL face database and the NORB dataset demonstrate that ELM-HLRF outperforms ELM-LRF, ELM and support vector machine in classification accuracy, and the presented data augmentation method improves the classification performance.

Image super-resolution aims at generating high-resolution images from low-resolution inputs. In this paper, we propose a novel learning-based and efficient image super-resolution approach called particle swarm optimization based selective ensemble (PSOSEN) of local receptive fields based extreme learning machine (ELM-LRF). ELM-LRF is locally connected ELM, which can directly process information including strong correlations such as images. PSOSEN is a selective ensemble used to optimize the output of ELM-LRF. This method constructs an end-to-end mapping of which the input is a single low-resolution image and the output is a high resolution image. Experiments show that our method is better in terms of accuracy and speed with different magnification factors compared to the state-of-the-art methods.

Fishes are famous for their ability to position themselves accurately even in turbulent flows. This ability is the result of the coordinated movement of fins which extend from the body. We have embarked on a research program designed to develop an agile and high efficient biologically inspired robotic fish based on the performance of hybrid mechanical fins. To accomplish this goal, a mechanical ray-like fin actuated by Shape Memory Alloy (SMA) is developed, which can realize both oscillatory locomotion and undulatory locomotion. We first give a brief introduction on the mechanical structure of our fin and then carry out theoretic analysis on force generation. Detailed information of these theoretical results is later revealed by Computational Fluid Dynamic (CFD), and is final validated by experiments. This robotic fin has potential application as a propulsor for future underwater vehicles in addition to being a valuable scientific instrument.

We exploit the criteria to optimize training set construction for the large-scale video semantic classification. Due to the large gap between low-level features and higher-level semantics, as well as the high diversity of video data, it is difficult to represent the prototypes of semantic concepts by a training set of limited size. In video semantic classification, most of the learning-based approaches require a large training set to achieve good generalization capacity, in which large amounts of labor-intensive manual labeling are ineluctable. However, it is observed that the generalization capacity of a classifier highly depends on the geometrical distribution of the training data rather than the size. We argue that a training set which includes most temporal and spatial distribution information of the whole data will achieve a good performance even if the size of training set is limited. In order to capture the geometrical distribution characteristics of a given video collection, we propose four metrics for constructing/selecting an optimal training set, including salience, temporal dispersiveness, spatial dispersiveness, and diversity. Furthermore, based on these metrics, we propose a set of optimization rules to capture the most distribution information of the whole data using a training set with a given size. Experimental results demonstrate these rules are effective for training set construction in video semantic classification, and significantly outperform random training set selection.