Search

Retinal blood vessel detection has important clinical value in the diagnosis and treatment of fundus diseases. However, due to the complexity and diversity of fundus image features, most retinal segmentation methods have some problems such as low performance of blood vessel segmentation, weak anti-nose interference, and sensitivity to lesions. Therefore, a pixel points classification method based on ensemble classified deep neural network is proposed. Firstly, different residual network modes are used to classify pixel points and get the vascular segmentation image. Secondly, through the ensemble earning method, the segmentation results of each model are processed to obtain the final retinal vascular segmentation image. The simulation results on STARE, DRIVE, and CHASE datasets show that the segmentation accuracy is 97.36%, 95.57%, 96.36%, the specificity 98.06%, 97. 76%, 97.84%, and the F-measure 84. 98%, 82. 25%, 79. 87%. The F-measure 0. 23%, 0. 54%, and 0. 59% high¬er than R2U_Net. [ABSTRACT FROM AUTHOR]

Improving the generalization ability and reducing the over-fitting risk is the research focus of deep convolutional neural networks. Occlusion is one of the critical factors affecting the generalization ability of convolutional neural networks. It is usually hoped that the models after complex training can have a good generalization for occlusion images. In order to reduce the over-fitting risk and improve the robustness of the model to random occlusion image recognition, this paper proposes an activation feature processing algorithm. During the training process, the input image is occluded by processing the maximum activation feature map of a convolutional layer, then the occluded new image is used as a new input to the network to go on training the model. The experimental results show that the proposed algorithm can improve the classification performance of multiple convolutional neural network models on different datasets and the trained models have excellent robustness to the identification of random occlusion images. [ABSTRACT FROM AUTHOR]

The problems existing in the water quality safety for interconnected river system network are summarized and the key techniques used in construction of the system are extracted, an overall system framework of intelligent control and emergency treatment of water quantity and water quality for the interconnected river system network based on the internet of things is put forward, which consists of real-time sensing layer, interconnection of water quantity and quality and information layer, smart decision layer. The realized functions in the system include the intelligent sensing, intelligent simulation, intelligent diagnosis, intelligent early warning, intelligent scheduling, intelligent disposal, intelligent control, et al. [ABSTRACT FROM AUTHOR]

The aim of the experiment was to investigate the effect of Chinese herbal compound additive on growth performance, serum biochemical indices and spleen IL-2 mRNA expression of Hy-line Brown egg chicks. A total of 960 one-day-old healthy Hy-line Brown egg chicks were randomly divided into two groups with five replicates per group and 96 chickens per replicate. Egg chicks in control group were fed with basal diet, and those in experimental group were fed with basal diet supplemented with 0.2% Chinese herbal compound additive. The experiment lasted for 42 d. The results showed that compared with control group, the final weight and average daily gain of egg chicks in experimental group were extremely increased (P<0.01), and the feed to gain ratio and mortality rate were extremely decreased (P<0.01). Compared with control group, keel length and tibia length of egg chicks in experimental group were extremely increased (P<0.01), and chest width and hip width were significantly increased (P<0.05). Compared with control group, the content of BUN and TC in serum of egg chicks in experimental group were extremely decreased (P<0.01), the activity of AST was significantly decreased (P<0.05), and the content of IFN-γ and IL-2 were extremely increased (P<0.01). Compared with control group, thymus index, bursa of Fabricius index and spleen IL-2 mRNA expression of egg chicks in experimental group were extremely increased (P<0.01). The study indicates that 0.2% Chinese herbal compound additive in the diet can improve the growth performance, improve the serum biochemical indexes, and promote the development of immune organs and the expression of IL-2 mRNA in spleen of Hy-line Brown egg chicks. [ABSTRACT FROM AUTHOR]

Aimed to the issues of limitation of packet transmission with low priority and single-hop environment of enhanced distribution channel access (EDCA) in 802. lie-based wireless Mesh networks (WMNs), this paper designed a cross-layer mechanism based on pure EDCA named CL-EDCA. By taking packet's end-to-end forwarding delay and hops bound requirement, CL-EDCA could dynamically adjust packet's priority level in each flow at network layer based on per access category delay(PACD) and queue length, which were gathered from link and mac layer, to effectively adjust the utilization of access category (AC) and EDCA. Simulation results show CL-EDCA can provide excellent end-to-end delay assurance while achieving much higher fairness of data transmission, compared to the pure EDCA scheme. [ABSTRACT FROM AUTHOR]

UF-growth algorithm generated a large number of tree nodes and branches, and it needed to constantly calculate the support of candidate data items. To solve these problems, this paper proposed a new algorithm named compressed UF-tree. The algorithm changed the condition of constructing tree. Once the data item in transaction matched the tree node in a branch, merged the data item to the branch. Otherwise, it created a new branch from the tree node. Finally, it saved TID of transaction in the leaf node. Building probability vector of single data item and calculating the support of candidate data items through TID and probability vector to mine frequent items. It proves that it's feasible and efficient according to experimental comparison and analysis. [ABSTRACT FROM AUTHOR]

In IEEE802. 11e-based wireless mesh networks, the parameters (eg., contention windows, transmission opportunity) used in 802. 11e's mechanism is statically assigned in order to guarantee the quality of service, which is more suitable to the communication nodes in one hop transmission with environment and not to the wireless multi-hop environment. In view of the problem that data flows with long delay bound may consume more channel time than data flows with short one so as to cause packet loss and networks performance degradation, a fair share scheme based on the residual packet delay bound, which is equally distributed among data packet, is proposed to alleviate the retransmission number and promote the fairness between data flows. Simulation results show that the proposed scheme is more effective than the original 802. 11e protocol. [ABSTRACT FROM AUTHOR]

The vibration equation is developed by semi-analytical finite element method. The mode-decoupling and Fourier transform are used to deal with the vibration equation, and the expression of dynamic stiffness is obtained. Considering the randomness of structural parameters, the random expression of dynamic stiffness is deduced by Taylor expansion and the random factor method. The influences of random factors on the means and variation coefficient of dynamic stiffness are discussed. The calculation resuits show that the standard deviation of elastic modulus random factor has a significant influence on the randomness of dynamic stiffness, and if all other factors are equal and the standard deviation of elastic modulus random factor is doubled, the decreasing ratios of all dynamic stiffness mean values in three minimum points are 46.7%, and are 27.4% and 3.6% for two maximum points in proper order. Furthermore, when the variability of structural parameters belongs to week degree, the variability of dynamic stiffness in extreme points belong to moderate degree, even strong degree. [ABSTRACT FROM AUTHOR]

Rolling bearings are an important part of problems that often occur in rotating machinery. Their fault conditions are complex and difficult to diagnose. Aiming at the problem that small sample learning has a large difference between the true feature value and the target feature value and the generalization ability is weak, this paper proposes a small sample learning model that combines semi-supervised variational autoencoder and LightGBM classification model. The Bayesian optimization and improvement algorithm based on Gaussian process is used to optimize the LightGBM hyperparameters, thus effectively solving the defects such as unstable performance, weak ability of extracting features, and overfitting in small sample learning. Comparative verification on the bearing experimental data set released by the Western Reserve University in the United States shows that the method has better diagnostic accuracy when facing small sample data space. [ABSTRACT FROM AUTHOR]

Aimed at the axial disturbed load to the head of supercavitating projectile at high speed motion underwater, its indefiniteness was considered, and the structure reliability of supercavitating projectile was analyzed on the basis of dynamic stability. The partial differential dynamic equation of the projectile's cut-off-cone structure was established, then it was transformed to a second-order ordinary differential Mathieu equation, and the numerical calculation for the dynamic stability of supercavitating structure was performed with Bolotin method, finally, the dynamic stability safety margin equations of supercavitating structure was given. The influences of length-diameter ratio, velocity and length of supercavitating projectile on the dynamic stability were analyzed. Based on those, the structural non-probabilistic reliability was analyzed and obtained, the effectiveness of the method was demonstrated by using some examples. [ABSTRACT FROM AUTHOR]