Attention-relation network for mobile phone screen defect classification via a few samples

How to use a few defect samples to complete the defect classification is a key challenge in the production of mobile phone screens.An attention-relation network for the mobile phone screen defect classification is proposed in this paper.The architecture of the attention-relation network contains two modules:a feature extract module and a feature metric module.Different from other few-shot models,an attention mechanism is applied to metric learning in our model to measure the distance between features,so as to pay attention to the correlation between features and suppress unwanted information.Besides,we combine dilated convolution and skip connection to extract more feature information for follow-up processing.We validate attention-relation network on the mobile phone screen defect dataset.The experimental results show that the classification accuracy of the attentionrelation network is 0.9486 under the 5-way 1-shot training strategy and 0.9039 under the 5-way 5-shot setting.It achieves the excellent effect of classification for mobile phone screen defects and outperforms with dominant advantages.

BARN:Behavior-Aware Relation Network for multi-label behavior detection in socially housed macaques

Quantification of behaviors in macaques provides crucial support for various scientific disciplines,including pharmacology,neuroscience,and ethology.Despite recent advancements in the analysis of macaque behavior,research on multi-label behavior detection in socially housed macaques,including consideration of interactions among them,remains scarce.Given the lack of relevant approaches and datasets,we developed the Behavior-Aware Relation Network(BARN)for multi-label behavior detection of socially housed macaques.Our approach models the relationship of behavioral similarity between macaques,guided by a behavior-aware module and novel behavior classifier,which is suitable for multi-label classification.We also constructed a behavior dataset of rhesus macaques using ordinary RGB cameras mounted outside their cages.The dataset included 65?913 labels for19 behaviors and 60?367 proposals,including identities and locations of the macaques.Experimental results showed that BARN significantly improved the baseline SlowFast network and outperformed existing relation networks.In conclusion,we successfully achieved multilabel behavior detection of socially housed macaques with both economic efficiency and high accuracy.

Graph Convolutional Networks Embedding Textual Structure Information for Relation Extraction

Deep neural network-based relational extraction research has made significant progress in recent years,andit provides data support for many natural language processing downstream tasks such as building knowledgegraph,sentiment analysis and question-answering systems.However,previous studies ignored much unusedstructural information in sentences that could enhance the performance of the relation extraction task.Moreover,most existing dependency-based models utilize self-attention to distinguish the importance of context,whichhardly deals withmultiple-structure information.To efficiently leverage multiple structure information,this paperproposes a dynamic structure attention mechanism model based on textual structure information,which deeplyintegrates word embedding,named entity recognition labels,part of speech,dependency tree and dependency typeinto a graph convolutional network.Specifically,our model extracts text features of different structures from theinput sentence.Textual Structure information Graph Convolutional Networks employs the dynamic structureattention mechanism to learn multi-structure attention,effectively distinguishing important contextual features invarious structural information.In addition,multi-structure weights are carefully designed as amergingmechanismin the different structure attention to dynamically adjust the final attention.This paper combines these featuresand trains a graph convolutional network for relation extraction.We experiment on supervised relation extractiondatasets including SemEval 2010 Task 8,TACRED,TACREV,and Re-TACED,the result significantly outperformsthe previous.

Genetic Relation Analysis on Ramie[Boehmeria nivea(L.)Gaud.]Inbred Lines by SRAP Markers

The objective of this article is to reveal the variations of ramie inbred lines in DNA level and discuss their molecular background to provide a theoretical basis for ramie cross breeding. In the present study, the genetic relationships among 33 inbred line accessions and two wild types that originated from China and Brazil were estimated using sequence-related amplified polymorphism （SRAP） markers. The results showed that 33 out of 81 primer combinations turned out to be polymorphic and 332 polymorphism bands were obtained. On the basis of the appearance of the markers, the genetic relationships were analyzed using unweighted pair-group method of arithmetic average cluster analysis （UPGMA）, and the genetic Jaccard similarity coefficients were calculated. The inbred-lines originating from China and Brazil formed a cluster suggesting a possibility that the Brazilian cultivars could have developed from cultivars introduced from China. Within ramie inbred-lines, the groupings also indicated that the greatest genetic relationship among cultivars was correlated to the region of origin of cultivars. The results provided the evidence that SRAP was an efficient approach, suitable for taxonomic analysis of ramie inbred lines, To the authors＇ knowledge, this is the first application of SRAP marker on the systematics of ramie inbred lines.

Function chain neural network prediction on heat transfer performance of oscillating heat pipe based on grey relational analysis

As for the factors affecting the heat transfer performance of complex and nonlinear oscillating heat pipe (OHP),grey relational analysis (GRA) was used to deal with the relationship between heat transfer rate of a looped copper-water OHP and charging ratio,inner diameter,inclination angel,heat input,number of turns,and the main influencing factors were defined.Then,forecasting model was obtained by using main influencing factors (such as charging ratio,interior diameter,and inclination angel) as the inputs of function chain neural network.The results show that the relative average error between the predicted and actual value is 4%,which illustrates that the function chain neural network can be applied to predict the performance of OHP accurately.

Relationship Between Heterosis and Parental Genetic Distance Based on Molecular Markers for Functional Genes Related to Yield Traits in Rice

The genetic distances among 18 cytoplasmic male sterile lines and 11 restorer lines were analyzed with molecular markers derived from yield-related functional genes. The correlation between parental genetic distance and heterosis was investigated by analyzing the performance of 47 combinations. The results showed that the genetic distance was significantly correlated with yield heterosis （r=0.29^＊）, but not significantly correlated with heterosis for other traits, such as number of effective panicles per plant, seed setting rate, 1000-grain weight, number of grains per panicle and theoretical yield. However, the correlation coefficient was so small that the parental genetic distance could not to be used to predict heterosis.

Selective kernel networks for weakly supervised relation extraction

The purpose of relation extraction is to identify the semantic relations between entities in sentences that contain two entities.Recently,many variants of the convolution neural network(CNN)have been introduced to relation extraction for the extracting of features--the quality of the neural network model directly affects the final quality of relation extraction.However,the traditional convolution network uses a fixed convolution kernel,so it is difficult to choose the size of the convolution kernel dynamically,which results in networks with weak representation ability.To address this,a novel CNN is designed with selective kernel networks and multigranularity.In the process of feature extraction,the model can adaptively select the size of the convolution kernel,that is,give more weight to the appropriate convolution kernel.It is then combined with multigranularity convolution to obtain more abundant semantic information.Finally,a new pooling method is designed to obtain more comprehensive information and improve model performance.Experimental results indicate that this method is effective without excessively deep network layers,and it also outperforms several competitive baseline methods.

A Knowledge-Enriched and Span-Based Network for Joint Entity and Relation Extraction

The joint extraction of entities and their relations from certain texts plays a significant role in most natural language processes.For entity and relation extraction in a specific domain,we propose a hybrid neural framework consisting of two parts:a span-based model and a graph-based model.The span-based model can tackle overlapping problems compared with BILOU methods,whereas the graph-based model treats relation prediction as graph classification.Our main contribution is to incorporate external lexical and syntactic knowledge of a specific domain,such as domain dictionaries and dependency structures from texts,into end-to-end neural models.We conducted extensive experiments on a Chinese military entity and relation extraction corpus.The results show that the proposed framework outperforms the baselines with better performance in terms of entity and relation prediction.The proposed method provides insight into problems with the joint extraction of entities and their relations.

Material component to non-linear relation between sediment yield and drainage network development:an flume experimental study

This paper examines the experimental study on influence of material component to non-linear relation between sediment yield and drainage network development completed in the Lab. The area of flume drainage system is 81.2 m2, the longitudinal gradient and cross section slope are from 0.0348 to 0.0775 and from 0.0115 to 0.038, respectively. Different model materials with a medium diameter of 0.021 mm, 0.076 mm and 0.066 mm cover three experiments each. An artificial rainfall equipment is a sprinkler-system composed of 7 downward nozzles, distributed by hexagon type and a given rainfall intensity is 35.56 mm/hr.cm2. Three experiments are designed by process-response principle at the beginning the ψ shaped small network is dug in the flume. Running time spans are 720 m, 1440 minutes and 540 minutes for Runs I, IV and VI, respectively. Three experiments show that the sediment yield processes are characterized by delaying with a vibration. During network development the energy of a drainage system is dissipated by two ways, of which one is increasing the number of channels (rill and gully), and the other one is enlarging the channel length. The fractal dimension of a drainage network is exactly an index of energy dissipation of a drainage morphological system. Change of this index with time is an unsymmetrical concave curve. Comparison of three experiments explains that the vibration and the delaying ratio of sediment yield processes increase with material coarsening, while the number of channel decreases. The length of channel enlarges with material fining. There exists non-linear relationship between fractal dimension and sediment yield with an unsymmetrical hyperbolic curve. The absolute value of delaying ratio of the curve reduces with time running and material fining. It is characterized by substitution of situation to time.

Relationships Between Fractal Road and Drainage Networks in Wuling Mountainous Area:Another Symmetric Understanding of Human-Environment Relations

Symmetrical relationships between humans and their environment have been referred to as an extension of symmetries in the human geographical system and have drawn great attention. This paper explored the symmetry between physical and human systems through fractal analysis of the road and drainage networks in Wuling mountainous area. We found that both the road and drainage networks reflect weak clustering distributions. The evolution of the road network shared a significant self-organizing composition, while the drainage network showed obvious double fraetal characteristics. The geometric fractal dimension of the road network was larger than that of the drainage network. In addition, when assigned a weight relating to hierarchy or length, neither the road network nor drainage network showed a fractal property. These findings indicated that the fractal evolution of the road network shared certain similarities with fractal distribution of the drainage network. The symmetry between the two systems resulted from an interactive process of destroying symmetry at the lower order and reconstructing symmetry at the higher order. The relationships between the fractal dimensions of the rural-urban road network, the drainage network andthe urban system indicated that the development of this area was to achieve the symmetrical isomorphism of physical-human geographical systems.

STRNet:Triple-stream Spatiotemporal Relation Network for Action Recognition

Learning comprehensive spatiotemporal features is crucial for human action recognition. Existing methods tend to model the spatiotemporal feature blocks in an integrate-separate-integrate form, such as appearance-and-relation network(ARTNet) and spatiotemporal and motion network(STM). However, with blocks stacking up, the rear part of the network has poor interpretability. To avoid this problem, we propose a novel architecture called spatial temporal relation network(STRNet), which can learn explicit information of appearance, motion and especially the temporal relation information. Specifically, our STRNet is constructed by three branches,which separates the features into 1) appearance pathway, to obtain spatial semantics, 2) motion pathway, to reinforce the spatiotemporal feature representation, and 3) relation pathway, to focus on capturing temporal relation details of successive frames and to explore long-term representation dependency. In addition, our STRNet does not just simply merge the multi-branch information, but we apply a flexible and effective strategy to fuse the complementary information from multiple pathways. We evaluate our network on four major action recognition benchmarks: Kinetics-400, UCF-101, HMDB-51, and Something-Something v1, demonstrating that the performance of our STRNet achieves the state-of-the-art result on the UCF-101 and HMDB-51 datasets, as well as a comparable accuracy with the state-of-the-art method on Something-Something v1 and Kinetics-400.

Electrocatalytic volcano relations:surface occupation effects and rational kinetic models

Electrocatalysis plays a vital role in technologies of energy and environment relevance,such as water electrolysis,fuel cells,synthesis of carbon and nitrogen-based fuels,etc.The volcano relations(VRs)are general and standard tools for predicting and understanding the activity trends of electrocatalysts.The modern electrocatalytic VRs are generally based on the kinetic models with the maximum free energy(△G^(0)_(max))of reaction steps as the rate-determining term(RDT),in which some important factors that crucially impact the reaction kinetics are missed,for examples,the surface structures and coverages of reaction intermediates and spectators,other free energy demanding steps than that associated with the △G^(0)_(max),and so on.In this perspective,we first give a brief introduction of the theoretical framework of current electrocatalytic VRs and the underlying problems in the oversimplifiedDG0max-based kinetic models,and then provide an account of our effort in constructing more rational VRs for electrocatalytic reactions.We introduce a new theoretical framework of electrocatalytic VRs based on kinetic model with the so-called energetic span(δE)serving as RDT.Since the surface-coverage effects and multiple free energy-demanding steps are considered,the VRs thus obtained show several new features such as strong potential dependence,asymmetric ascending and descending branches,relatively flat tops,and so on.The effectiveness of theδE-based VRs is verified for hydrogen and oxygen electrocatalytic reactions.Finally,research directions to further rationalize the electrocatalytic VRs are discussed.

Neural Network Model for the Constitutive Relations of Soil

The soil constitutive relation is one of the important issues in soil mechanics. It is very difficult to establish mathematical models because of the complexity of soil mechanical behavior. We propose a new method of neural network analysis for establishing soil constitutive models. Based on triaxial experiments of sand in the laboratory, the nonlinear constitutive models of sand expressed by the neural network were set up. In comparison with Duncan\|Chang's model, the neural network method for sand modeling has been proved to be more convenient, accurate and it has a strong fault\|tolerance function.

Spatial Interaction Network Analysis of Crude Oil Trade Relations between Countries along the Belt and Road

Based on the theories and methods of complex network,crude oil trade flows between countries along the Belt and Road(B&R,hereafter)are inserted into the Geo-space of B&R and form a spatial interaction network which takes the countries as nodes and takes the trade relations as edges.The networked mining and evolution analysis can provide important references for the research on trade relations among the B&R countries and the formulation of trade policy.This paper researches and discusses the construction,statistical analysis,top networks and stability of the crude oil trade network between the B&R countries from 2001 to 2020 from the perspectives of Geo-Computation for Social Sciences(GCSS)and spatial interaction.Firstly,evolutions of out-degree,in-degree,out-strength and in-strength of the top 10 countries in the crude oil trade network are computed and analyzed.Secondly,the top network method is used to explore the evolution characteristics of hierarchical structures.And finally,the sequential evolution characteristics of the crude oil trade network stability are analyzed utilizing the network stability measure method based on the trade relationship autocorrelation function.The analysis results show that Russia has the largest out-degree and out-strength,and China has the largest in-degree and in-strength.The crude oil trade volume of the top 10 import and export networks between 2001—2020 accounts for over 90%of the total trade volume of the crude oil trade network,and the proportion remains relatively stable.However,the stability of the network showed strong fluctuations in 2009,2012 and 2014,which may be closely related to major international events in these years,which could furtherly be used to build a correlation model between network volatility and major events.This paper explores how to construct and analyze the spatial interaction network of crude oil trade and can provide references for trade relations research and trade policy formulation of B&R countries.

Relationship of Internet Addiction with Depression, Loneliness and Health Related Lifestyle among University Students

Internet addiction （IA） is a newly emerged clinical disorder and it has negative effects on physical and mental health. University students are the most vulnerable group for IA, The aim of the present study was to determine the relationship of IA with depression, loneliness and health related lifestyle among university students. Cross-sectional survey was conducted by enrolling 175 students of Faculty of Allied Health Sciences, University of Peradeniya. Internet addiction test （IAT） was used to assess the level of IA. Depression, loneliness, and health related lifestyle were assessed using Peradeniya depression scale （PDS）, University of California at Los Angeles （UCLA） loneliness scale and health practice score （HPS） respectively. T-test and ANOVA were conducted to examine the differences; and correlation and regression analyses were used to examine the relationships between variables. Overall, 40.6% of students were placed in IA group. Generally 28.6% of students had mild and 12.0% had moderate addiction. No case of severe IA was seen. There were 20.6% of students in depressive state and 17.1% of students had poor HPS. The average score that the student got from loneliness scale was 23.42. There was a positive significant correlation between IA and both depression and loneliness. Moreover, a negative significant correlation found between IA and health related lifestyle. Male students had higher IA scores than female students. The study results are considered to develop preventive interventions and treatment strategies.

Joint Self-Attention Based Neural Networks for Semantic Relation Extraction

Relation extraction is an important task in NLP community.However,some models often fail in capturing Long-distance dependence on semantics,and the interaction between semantics of two entities is ignored.In this paper,we propose a novel neural network model for semantic relation classification called joint self-attention bi-LSTM(SA-Bi-LSTM)to model the internal structure of the sentence to obtain the importance of each word of the sentence without relying on additional information,and capture Long-distance dependence on semantics.We conduct experiments using the SemEval-2010 Task 8 dataset.Extensive experiments and the results demonstrated that the proposed method is effective against relation classification,which can obtain state-ofthe-art classification accuracy just with minimal feature engineering.

PDRK:A General Kinetic Dispersion Relation Solver for Magnetized Plasma

A general,fast,and effective approach is developed for numerical calculation of kinetic plasma linear dispersion relations.The plasma dispersion function is approximated by J-pole expansion.Subsequently,the dispersion relation is transformed to a standard matrix eigenvalue problem of an equivalent linear system.Numerical solutions for the least damped or fastest growing modes using an 8-pole expansion are generally accurate;more strongly damped modes are less accurate,but are less likely to be of physical interest.In contrast to conventional approaches,such as Newton＇s iterative method,this approach can give either all the solutions in the system or a few solutions around the initial guess.It is also free from convergence problems.The approach is demonstrated for electrostatic dispersion equations with one-dimensional and twodimensional wavevectors,and for electromagnetic kinetic magnetized plasma dispersion relation for bi-Maxwellian distribution with relative parallel velocity flows between species.

Uncertain Relation Suited to Overfitting of BP Neural Network

A general uncertainty relation between the change of weighted value which represents learning ability and the discrimination error of unlearning sample sets which represents generalization ability is revealed in the modeling of back propagation (BP) neural network. Tests of numerical simulation for multitype of complicated functions are carried out to determine the value distribution (1×10?5~5×10?4) of overfitting parameter in the uncertainty relation. Based on the uncertainty relation, the overfitting in the training process of given sample sets using BP neural network can be judged.

Scattering Relations for Two-Dimensional Electromagnetic Waves in Chiral Media

We consider the scattering of time-harmonic plane waves by an infinitely long penetrable chiral cylinder. The electromagnetic scattering problem is reduced to a transmission problem for a system of two-dimensional Helmholtz equations. We prove the classical reciprocity principle, a general scattering theorem and an optical theorem in R2. Using Herglotz wave functions we define the corresponding far field operator. Applying the general scattering theorem useful relations are proved for the reconstruction of the scatterer. We also prove that for real chirality measure of the penetrable scatterer the far field operator has a countable number of eigenvalues which lie on a circle.

Tunable dispersion relations manipulated by strain in skyrmion-based magnonic crystals

We theoretically investigate the propagation characteristics of spin waves in skyrmion-based magnonic crystals. It is found that the dispersion relation can be manipulated by strains through magneto-elastic coupling. Especially, the allowed bands and forbidden bands in dispersion relations shift to higher frequency with strain changing from compressive to tensile,while shifting to lower frequency with strain changing from tensile to compressive. We also confirm that the spin wave with specific frequency can pass the magnonic crystal or be blocked by tuning the strains. The result provides an advanced platform for studying the tunable skyrmion-based spin wave devices.