Tech Science Press
Not a member yet
3972 research outputs found
Sort by
Researching the Link Between the Geometric and Rènyi Discord for Special Canonical Initial States Based on Neural Network Method
Full text linkQuantum correlation which is different to the entanglement and classical correlation plays important role in quantum information field. In our setup, neural network method is adopted to simulate the link between the Rènyi discord (α = 2) and the geometric discord (Bures distance) for special canonical initial states in order to show the consistency of physical results for different quantification methods. Our results are useful for studying the differences and commonalities of different quantizing methods of quantum correlation
A Method of Obtaining Catchment Basins with Contour Lines for Foam Image Segmentation
Full text linkFoam image segmentation, represented by watershed algorithm, is wildly used in the extraction of bubble morphology features. H-minima transformation was proved to be effective in locating the catchment basins in the traditional watershed segmentation method. To further improve the accuracy of watershed segmentation, method of top-bottom-cap filters and method of morphological reconstruction were implied to marking the catchment basins. In this paper, instead of H-minima transformation, a method of contour lines is specially proposed to obtain the catchment basins for foam image segmentation by using top-bottom-cap filters and less morphological reconstruction. Experimental results in foam segmentation show that the proposed method is equally accurate but more efficient than the method of H-minima plus morphological reconstruction, and equally efficient but more accurate than the method of H-minima plus top-bottom-cap filters
Distant Supervised Relation Extraction with Cost-Sensitive Loss
Full text linkRecently, many researchers have concentrated on distant supervision relation extraction (DSRE). DSRE has solved the problem of the lack of data for supervised learning, however, the data automatically labeled by DSRE has a serious problem, which is class imbalance. The data from the majority class obviously dominates the dataset, in this case, most neural network classifiers will have a strong bias towards the majority class, so they cannot correctly classify the minority class. Studies have shown that the degree of separability between classes greatly determines the performance of imbalanced data. Therefore, in this paper we propose a novel model, which combines class-to-class separability and cost-sensitive learning to adjust the maximum reachable cost of misclassification, thus improving the performance of imbalanced data sets under distant supervision. Experiments have shown that our method is more effective for DSRE than baseline methods
A DDoS Attack Situation Assessment Method via Optimized Cloud Model Based on Influence Function
Full text linkThe existing network security situation assessment methods cannot effectively assess the Distributed denial-of-service (DDoS) attack situation. In order to solve these problems, we propose a DDoS attack situation assessment method via optimized cloud model based on influence function. Firstly, according to the state change characteristics of the IP addresses which are accessed by new and old user respectively, this paper defines a fusion feature value. Then, based on this value, we establish a V-Support Vector Machines (V-SVM) classification model to analyze network flow for identifying DDoS attacks. Secondly, according to the change of new and old IP addresses, we propose three evaluation indexes. Furthermore, we propose index weight calculation algorithm to measure the importance of different indexes. According to the fusion index, which is optimized by the weighted algorithm, we define the Risk Degree (RD) and calculate the RD value of each network node. Then we obtain the situation information of the whole network according to the RD values, which are from each network nodes with different weights. Finally, the whole situation information is classified via cloud model to quantitatively assess the DDoS attack situation. The experimental results show that our method can not only improve the detection rate and reduce the missing rate of DDoS attacks, but also access the DDoS attack situation effectively. This method is more accurate and flexible than the existing methods
Optical design and performance comparison of various hyperspectral imagers based on Fery prisms
Full text linkThe paraxial ray-tracing model of Fery prism is illustrated in this paper, and the three-order aberration coefficients are calculated. According to the solutions for minimal aberrations, four types of imaging spectrometers are designed based on Fery prism, accommodating for different requirements. The image quality is evaluated to ensure that MTFs are larger than 0.6 at Naquist frequencies, and the spectral resolutions are all higher than 5 nm. The advantages of these imaging spectrometers are analyzed by comparison of the volume, spectral resolution and field of view. The potential competence of Fery prism in hyperspectral imaging is indicated since the field of view and the volume have a direct proportional function. One typical system of these designs is manufactured and assembled afterwards to verify the simulation data
Convolution Neural Networks and Support Vector Machines for Automatic Segmentation of Intracoronary Optical Coherence Tomography
Full text linkCardiovascular diseases are closely associated with deteriorating atherosclerotic plaques. Optical coherence tomography (OCT) is a recently developed intravascular imaging technique with high resolution approximately 10 microns and could provide accurate quantification of coronary plaque morphology. However, tissue segmentation of OCT images in clinic is still mainly performed manually by physicians which is time consuming and subjective. To overcome these limitations, two automatic segmentation methods for intracoronary OCT image based on support vector machine (SVM) and convolutional neural network (CNN) were performed to identify the plaque region and characterize plaque components. In vivo IVUS and OCT coronary plaque data from 5 patients were acquired at Emory University with patient’s consent obtained. Seventy-seven matched IVUS and OCT slices with good image quality and lipid cores were selected for this study. Manual OCT segmentation was performed by experts using virtual histology IVUS as guidance, and used as gold standard in the automatic segmentations. The overall classification accuracy based on CNN method achieved 95.8%, and the accuracy based on SVM was 71.9%. The CNN-based segmentation method can better characterize plaque compositions on OCT images and greatly reduce the time spent by doctors in segmenting and identifying plaques
Influence of Sugarcane Bagasse Fiber Size on Biodegradable Composites of Thermoplastic Starch
Full text linkAlthough thermoplastic starch (TPS) is biodegradable, its low mechanical resistance limits its wide application. Sugarcane bagasse (SB) fibers can be used as reinforcement in TPS matrix composites, but the influence of fiber size on the properties of the composite is still unknown. In this study, TPS composites reinforced with SB short fibers of four sizes were processed and characterized in order to analyze the influence of fiber size on the mechanical properties of the TPS/SB composite. It was observed that the interaction between fiber and matrix was good and optimized when the fibers are sifted in sieves between 30 and 50 mesh, obtaining fibers with average length of 1569 ± 640 μm and average diameter of 646 ± 166 μm. For these composites, increases of more than 660% in the modulus and more than 100% in the maximum tension were verified when compared to the pure TPS
Analysis and Test on Influence Factors of Dew Drop Condensation in Dew Point Hygrometer
Full text linkThe condensation process of dew droplets is influenced by many factors. A dew point condensation image observation system was built to improve the response speed of dew point detector under different measuring conditions. The basic mechanism of dew drop condensation growth was studied and the influence of various factors on the dew drop growth rate were analyzed. And the accuracy of the influence results was verified based on the improved Hough transform circle detection. The results show that the growth rate of dew droplets is affected by ambient temperature, dew point temperature, mirror temperature and air velocity. The observed variation of the average radius of dew droplets is consistent with the theoretical calculations. The maximum radius error is less than 4 μm, the initial error is larger, and the error oscillates in the middle and late stages of condensation. The establishment of condensation mechanism is helpful to solve the problem in fast determination of dew point temperature under the cold start of dew point meter, and to improve the response speed
Tibetan Multi-Dialect Speech Recognition Using Latent Regression Bayesian Network and End-To-End Mode
Full text linkWe proposed a method using latent regression Bayesian network (LRBN) to extract the shared speech feature for the input of end-to-end speech recognition model. The structure of LRBN is compact and its parameter learning is fast. Compared with Convolutional Neural Network, it has a simpler and understood structure and less parameters to learn. Experimental results show that the advantage of hybrid LRBN/Bidirectional Long Short-Term Memory-Connectionist Temporal Classification architecture for Tibetan multi-dialect speech recognition, and demonstrate the LRBN is helpful to differentiate among multiple language speech sets
Deep Learning Trackers Review and Challenge
No full textRecently, deep learning has achieved great success in visual tracking. The goal of this paper is to review the state-of-the-art tracking methods based on deep learning. First, we categorize the existing deep learning based trackers into three classes according to network structure, network function and network training. For each categorize, we analyze papers in different categories. Then, we conduct extensive experiments to compare the representative methods on the popular OTB-100, TC-128 and VOT2015 benchmarks. Based on our observations. We conclude that: (1) The usage of the convolutional neural network (CNN) model could significantly improve the tracking performance. (2) The trackers with deep features perform much better than those with low-level hand-crafted features. (3) Deep features from different convolutional layers have different characteristics and the effective combination of them usually results in a more robust tracker. (4) The deep visual trackers using end-to-end networks usually perform better than the trackers merely using feature extraction networks. (5) For visual tracking, the most suitable network training method is to per-train networks with video information and online fine-tune them with subsequent observations. Finally, we summarize our manuscript and highlight our insights, and point out the further trends for deep visual tracking