Search

This paper proposed a natural language sentence matching method that combined high-level and low-level semantic information to solve the problems about current natural language sentence matching method fail to integrate common semantic information and it is difficult to capture deep-semantic information. First of all, the method used pre-trained word vector GloVe and character-level word vector to obtained the word embedding representation of sentence P and sentence Q. Secondly, this paper encode red P and Q with bidirectional LSTM, then it contained low-level semantic information through preliminary fusion of P and Q. Thirdly, this paper calculated bidirectional attention between P and Q, then spliced them together to get semantic representation, afterwards it calculated its self-attention to obtained high-level semantic information. Finally, this paper used a heuristic fusion function to fuse the low-level semantic information with the high-level semantic information to obtain the final semantic representation, and it used a convolutional neural network to prediction answers. This paper evaluated the proposed model on two tasks, such as recognition textual entailment, paraphrase recognition. This paper conducted experiments on the SNLI dataset and the Quora dataset. The results show that the accuracy of the proposed algorithm on the SNLI test set is 87. 1%, and the accuracy of the Quora test set is 86. 8%, which verifies the effectiveness of the algorithm in the task of natural language sentence matching. [ABSTRACT FROM AUTHOR]

Facial local detail information plays an important role in facial expression recognition（FER）.However, most of the existing methods only focus on the high-level semantic information of facial expressions, while ignoring the fine-grained information of local facial regions. To solve this problem, this paper proposed a deep convolutional neural network fusing local feature and two-stage attention weight learning（FLF-TAWL）,which could adaptively capture important facial regions to improve the effectiveness of facial expression recognition. The FLF-TAWL model was composed of a dual-branch framework, one branch extracted local features from image blocks, and the other branch extracted global features from the entire expression image. Firstly, this paper proposed a two-stage attention weight learning strategy. In the first stage, it roughly learned the importance weights of global and local features, in the second stage, it further refined the attention weight, and fused the local and global features. Secondly, the model used a region-biased loss function to encourage the most important regions to obtain higher attention weights. Finally, this paper carried out extensive experiments on FERPlus, Cohn-Kanada（CK+） and JAFFE datasets to obtain accuracy rates of 90.92%,98.90% and 97.39% respectively. The experimental results verify the effectiveness and feasibility of the FLF-TAWL model. [ABSTRACT FROM AUTHOR]

Multi-agent reinforcement learning methods have been made great progress in simulation, game, recommendation systems and so on. However, the complex problems in the real word bring great difficulties for reinforcement learning, such as many useless explorations, slow converging speed and poor performance of the learning. This paper studied the problem of multi-agent reinforcement learning with embedded rules and proposed a method to combine rules and the learning method based on an iterative training mechanism. This method designed a multi-agent reinforcement learning method with embedded rules, and a rule selection model. This paper introduced an iterative training mechanism to combine the two methods together.The proposed method could decide whether to use the result of a reinforcement learning or the result of a rule based on the real-time game state. It could effectively solve the problem that which rule should be selected and when it would be used. Finally, it made an experiment on a multi-agent combat platform which was published by the China Electronics Technology Group. By fighting with the built-in opponent in the platform, it found that the method with rules could achieve 60% win rate with 14 thousand rounds while achieve 50% win rate with 17 thousand rounds for the method without rules. The results show that the proposed method can effectively improve the converging speed and the performance of multi-agent reinforcement learning. [ABSTRACT FROM AUTHOR]

Intelligent question answering based on reading comprehension refers to letting computers read and comprehend texts like humans, extracts the text information and answers corresponding questions. The pre-training model RoBERTa-wwm-ext uses the extracted original fragments as the answers to the questions, but this method can’t solve the two situations that the answer fragments don’t exist in the original text or need to reply to the original text after summarizing. The pre-training model is used for generative model training, which can solve the problems that need to summarize the original text to a certain extent. Therefore, this paper improved the method of only using RoBERTa-wwm-ext model to extract answers. On this basis, it integrated the generative question answering model based on RAG model to answer questions that could not be handled by Roberta-wwm-ext and other extraction models. At the same time, this paper absorbed the advantages of PGN model, improved RAG model, and obtained RPGN sub model, which could make better use of reading and understanding articles to generate reasonable answers. Therefore, this paper proposed a fusion model of RPR（RAG, PGN, RoBERTa-wwm-ext）,which could be used to deal with both extractive question task and generative question answering task at the same time. [ABSTRACT FROM AUTHOR]

Due to the lack of sufficient co-occurrence information in short text, weak connections between words, and it is difficult to obtain subject words, which leads to the need to manually label a large number of training samples for short text classification, and the problems of sparse features and dimension explosion. This paper proposed a word symbiotic short text classification model based on attention mechanism and label graph（WGA-BERT）.Firstly, this paper used the pretrained BERT model to calculate the context aware text representation, and used WNTM to model the potential word group distribution of each word to obtain the topic expansion feature vector. Secondly, this paper used a tag graph construction method to capture the structure and relevance of subject words. Finally, this paper used an attention mechanism to establish the relationship between subject words and between subject words and text, which solved the problems of data sparsity and subject text heterogeneity. The experimental results show that the WGA-BERT model improves the classification accuracy by an average of 3% compared with the traditional machine learning model. [ABSTRACT FROM AUTHOR]

In order to improve the search ability of AGV in complex unknown environment, this paper proposed an improved DDPG. It constructed the empirical playback matrix and double-layer network structure to improve the convergence speed of the algorithm ε-greedy in the green search strategy, to solve the local optimization problem of AGV in selecting the optimal action. To solve the problem of slow training speed of deep neural network, it applied priority sampling to depth deterministic strategy gradient algorithm. In order to solve the problem of high complexity of ordinary priority sampling, this paper proposed a small batch priority sampling method to train the network. In order to verify the effectiveness of the method, it used the grid method to model and compare the simulation experiments in different complex environments. It compared the loss function, iteration times and return value of different algorithms. The experimental results show that compared with the original algorithm, the improved algorithm reduces the loss function, reduces the number of iterations and increases the return value, which verifies the effectiveness of the algorithm. At the same time, it provides a new idea for AGV to complete the planning task more safely and quickly in complex environment. [ABSTRACT FROM AUTHOR]

Copyright of Application Research of Computers / Jisuanji Yingyong Yanjiu is the property of Application Research of Computers Edition and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Application Research of Computers / Jisuanji Yingyong Yanjiu is the property of Application Research of Computers Edition and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

When recommending items to groups on social networks, most of the existing studies use the complete preferences of group members and some aggregation strategies to generate group recommendation. However, in practice, a user may join a group just because of a part of his/her complete preferences related to the group. In this way, using the complete preferences of group members for recommendation will bring a lot of noise, and damage the effect of recommendation. This paper proposed the HINGR_GT method to solve the above problem. It constructed a group oriented heterogeneous social network sub graph for each group, and excluded members' group independent preferences in advance. On this basis, it used the heterogeneous network representation learning method to generate item recommendation for groups. In order to verify the effectiveness of the proposed method, this paper carried out experiments on Douban Movie dataset, and the results show that the proposed method achieves better results in all evaluation metrics. [ABSTRACT FROM AUTHOR]

Aiming at the problems of parallel support vector machine (SVM) algorithm in big data environment such as redundant data sensitivity, difficulty in parameter selection, and low parallelization efficiency, this paper proposed a parallel SVM algorithm using Relief and bacterial foraging optimization (BFO) algorithm based on MapReduce (RBFO-PSVM). Firstly, the algorithm designed a feature weight calculation strategy (MI-Relief), which used mutual information to improve the weight calculation function of Relief algorithm to eliminate redundant features in the data set and effectively reduce redundant data to support parallelism. Secondly, this paper proposed a hybrid BFO algorithm based on MapReduce (MR-HBFO), which selected the optimal parameters of SVM in parallel, and solved the problem of difficult selection of SVM parameters. Finally, it proposed the kernel clustering strategy (KCS) to reduce the size of the data set involved in parallel training, and proposed a cross-fusion cascaded parallel SVM (CFCPSVM) to improve the cascade SVM (CSVM) feedback mechanism. It trained SVM by combining with the Map Reduce programming framework, and this improved the parallelization efficiency of parallel SVM. Experiments show that the RBFO-PSVM algorithm has a better classification effect on large data sets and is more suitable for large data environments. [ABSTRACT FROM AUTHOR]

Address matching is one of the key technologies in the investigation and analysis of origin-destination in hazardous materials transportation. In order to help solving the problem of low accuracy of complex non-standard address matching for hazardous materials road transportation, this paper adopted self-expanding knowledge graph of address data, proposed a self-expanding Chinese word segmentation method and an address matching mechanism based on full-text index. Besides, this paper incorporated a weighted Pinyin full-text search mechanism to improve the matching accuracy for misspelling addresses. The online interface of geographic resolving and semi-supervised matching mechanism made the proposed address matching system a complete one. The matching results of the address data of the electronic waybill for the hazardous materials transportation show a high accuracy and high precision. The accuracy is over 94.6% for random address dataset and 67.5% for difficult-to-parse address dataset, both are much higher than the general matching methods and geographic search methods in terms of the accuracy and precision of address matching. [ABSTRACT FROM AUTHOR]

In order to solve the problem of premature convergence and falling into local optimum in biogeography-based optimization algorithm, this paper proposed an improved biogeography-based optimization algorithm introducing species evolution (SEBBO). The algorithm divided all habitats into three regions according to the number of species, established a co-evolution relationship, and reasonably adopted interval invasion and regional cooperation/competition strategies to meet diversity while avoided premature convergence. It defined two mutation strategies of species change and species development, and designed the dual-strategy cooperative mutation operator to solve the damage of the mutation operator to the better solution. Then, compared with standard BBO and other related improved algorithms on 8 benchmark functions from CEC2017, SEBBO algorithm is superior to BBO and other BBO variants in terms of algorithm performance and stability, and local optimal values is not easily limit the algorithm. Finally, this paper applied the SEBBO algorithm to flexible Job-Shop scheduling with maximum completion time as the goal to test its practical application value. Experiments show that the algorithm has a certain effectiveness in solving FJSP. [ABSTRACT FROM AUTHOR]

To solve the problem of insufficient real-time performance of existing optical flow estimation methods, this paper proposed a lightweight improved PSC-NET model with deep separable convolution ( depth pyramid, warping and cost volume, DS-PWC). One of the improvements was decoupling the conventional two-dimensional convolutional network layer into a deep separable convolutional layer. The other one was adding a weight coefficient based on the number of layers in the pyramid layer by DS-PWC, which greatly reduced the number of model parameters in the network structure without loss of precision. In addition, in the training process, the paper applied data enhancement technologies such as I+ ORE to further improve the generalization ability of estimation and prediction results. The experimental results show that the DS-PWC model was tested in the dataset and the operating efficiency reaches about 58 fps while maintaining the quality. To verify the effectiveness of the algorithm, this paper carried out the ablation experiments of model structure and data enhancement. The results verify the effectiveness of the DS-PWC model. [ABSTRACT FROM AUTHOR]

Poor text reading ability and inadequate visual reasoning were the main reasons for the insufficient effect of existing visual question answering models. To solve the above problems, this paper designed a MRGNN model. It used various forms of information in images to help understanding the scene text content, preprocessed the scene text image into the visual object graph and text graph respectively, and filtered the redundant information in the question self-attention module. It used an aggregator with attention to perfect the node features between subgraphs and fuse different modality information. The updated nodes used the context information of different modules to provide a better function for answering module. This paper verified the validity of MRGNN model on ST-VQA and Text VQA datasets. The experimental results show that MRGNN model achieves good results compared with some classical models for this task. [ABSTRACT FROM AUTHOR]

Gait recognition systems have shown great potentials in the field of biometric recognition. However, the accuracy of gait recognition is easily affected by a large view angle. In order to solve this problem, this paper proposed an improved model two-channel cycle consistency generative adversarial network( TCC-GAN). The network could transform the gait energy image from arbitrary views to the target view and then performed identity recognition. Besides reconstruction loss, it introduced view classification and identity preserving loss to guide the generator to produce gait images of the target view and keep identity information simultaneously. In order to avoid the model collapse problem that might exist and ensure that each input and output were mapped in a meaningful way, this paper also utilized the recently proposed cycle consistency loss, which could facilitate that the target image and the source image under the target view had the same feature distribution as possible. The experimental results on the datasets CASIA-B and OU-MVLP indicate that the proposed TCC-GAN model can obtain higher accuracy than other state-of-the-art GAN-based cross-view gait recognition models. [ABSTRACT FROM AUTHOR]

In order to solve the problem that it was difficult to effectively coordinate the maintenance of diagnosis and treatment equipment and inpatient waiting, this paper established a joint optimization model. Firstly, assuming that the equipment had discrete deterioration state, it modeled the deterioration process of the equipment as a continuous time Markov chain. Secondly, considering the different functional frequency requirements of patients for diagnosis and treatment equipment and the impact of different deterioration states on patients' treatment time and cost, it established the equipment maintenance and patient scheduling model with patients ' treatment order, inspection strategy and repair strategy as decision variables. Finally, this paper used the improved non dominated sorting genetic algorithm to solve the multi-objective problem. The experimental results verify the effectiveness of the joint optimization model of equipment maintenance and patient scheduling. The improved algorithm improves the overall and local search ability, and has robustness. [ABSTRACT FROM AUTHOR]

In order to improve the safety and smoothness of flight, and solve the problems of excessive turning angle and too many search path nodes in the traditional A• algorithm, this paper proposed a synchronous bi-directional A• search algorithm based on sector field expansion. Firstly, this paper extended the hazardous area boundaries based on the raster map method. Secondly, it designed the A• algorithm based on synchronous bi-directional search, which dynamically defined the target nodes for forward and reverse search. Aiming at the problem of limiting search angles, this paper proposed a sector expansion strategy in 5 x 5 domain, and designed an evaluation function with double weight parameters to reduce the generation of redundant points. Finally, in order to verify the effectiveness of the improved algorithm, this paper selected square and irregular shape hazard areas for simulation. The results show that the improved algorithm searches for a smoother path. Compared with the results of the traditional bi-directional A• algorithm, the length of its search path reduce by 1. 65% and 13. 16%, and the number of its search path nodes reduce by 42. 6% and 46. 81 %, respectively, under different shapes of hazard areas, which has a strong search efficiency. [ABSTRACT FROM AUTHOR]

The location problem of electric vehicle charging station is a hot issue in the current society, its essence is the classical NP-hard problem in combinatorial optimization. This paper studied the location of charging station based on the minimum opening cost. Firstly, this paper established the mathematical model of the problem, then studied the mathematical properties of the problem and gave the corresponding proof. It used these properties to reduce the scale of the problem, so as to reduce the difficulty of solving the problem. Then this paper designed the upper and lower bound sub algorithm and the reduced order sub algorithm. Based on these sub algorithms, this paper proposed the branch and bound algorithm which could quickly reduce the size of the problem and obtain the optimal solution, which reduced the time complexity and pruned the solution space to speed up the solution speed. Finally,it gave an example to illustrate the principle and implementation of the algorithm. [ABSTRACT FROM AUTHOR]

In order to solve the problem of collision free path planning and system efficiency improvement of multi AGV in dynamic unstable environment, this paper proposed a collision free path planning method based on time window for AGV. Firstly, this paper established a multi-AGV collision avoidance model, combined with the time window model, and divided the collisionfree path planning of multiple AGV into two stages: pre-planning and real-time planning. In the pre-planning stage, this paper calculated the conflict-free time window of multiple AGV and maximized the circulation of AGV in the system. In the real-time planning stage, this paper used the method of changing the AGV' s occupancy priority on the collision avoidance model and local re-planning to dynamically avoid collisions. Finally, a simulation experiment with a smart warehouse as an application case proves that the algorithm can effectively avoid the collision of multiple AGV, increase the circulation of AGV, and has good robustness and flexibility in a dynamic environment. [ABSTRACT FROM AUTHOR]

The sentence level extraction method is insufficient to solve the problem of Chinese event element dispersion. To solve this problem, this paper proposed a document level event extraction method based on context fusion. Firstly, the paper divided the document into paragraphs, and used bidirectional long and short memory network to extract sequence features of paragraphs. Secondly, the method used self-attention mechanism to capture the interaction information of paragraph context. Then the method combined the document sequence features with the interaction information to update the semantic representation. Finally, the method used sequence annotation to extract event elements and match event types. Compared with other event extraction methods on the same Chinese data set, the experimental results show that this method can effectively extract scattered event elements from documents, and improve the extraction performance of the model. [ABSTRACT FROM AUTHOR]

Copyright of Application Research of Computers / Jisuanji Yingyong Yanjiu is the property of Application Research of Computers Edition and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This paper proposed a self-encoding salient region extraction algorithm based on multi-layer convolutional to solve the problem that the existing convolutional neural network wasn’t sufficient in extracting image features, which led to the poor effect of salient extraction. When convolutional networks were used to extract image features, the shallow convolution features generally extracted the detailed features of the image, such as color, texture and location features, and the deep convolution features were generally referred to the semantic features of the image. On the coding layer, this algorithm downsampled the shallow convolution features and fused them into the deep convolution features, and upsampled the deep convolution features to the shallow convolution feature. Experiments show that this method can greatly improve the coding quality. In the decoding, it also fused the convolutional features by encoding to obtain the lost information in the decoding and also obtained a better decoded image. In addition, this paper designed a layer-by-layer supervision method to guide the training of the decoding layer, which used the standard region extraction map by down-sampling as the standard image for supervision training. The experimental results show that this algorithm can increase the F-measure by 0.071 on the basis of PAGRN, and reduce the MEA by 0.031 on average. [ABSTRACT FROM AUTHOR]

In order to solve the problem of extracting temporal, spatial and channel features, this paper proposed an ADGCN for skeleton action recognition. At first, it modeled the skeleton data and fed the movement of joints, bones and joints and bones information into the multi-stream framework of a single stream. And then it transferred the input data to the directed graph convolution network to extract the dependencies between joints and bones, after that, used the space-temporal channel attention network（STCN） to strengthen the key joints in each layer of network temporal, space and channels of information. The last, it calculated the accuracy of action recognition by the weighted average of the information of four streams, output action prediction results. This model trained and verified in two large data sets NTU-RGB+D and Kinectics-Skeleton. Compared with baseline DGNN（directed graph neural network）, the accuracy of this paper on two cross subsets CS and CV is improved by 2.43% and 1.2%, on NTU-RGB+D dataset. On Kinectics-skeleton dataset, the accuracy of top1 and top5 is increased by 0.7% and 0.9%, respectively. The proposed ADGCN can effectively enhance the performance of skeleton action recognition, and the effect is improved on both large data sets. [ABSTRACT FROM AUTHOR]

At present, researches on short-term load forecasting mainly focus on the analysis of influencing factors and the optimization of the model, and there are few researches on the robustness of the model. This paper took the extreme learning machine as the research object, conducted in-depth research on the robustness of the model, and applied it to the short-term load forecasting problem. The robustness of the ELM model was affected by the loss function. The current ELM model had poor robustness and low prediction accuracy when dealing with the samples containing outliers. To solve this problem, this paper proposed a new loss function based on p-order maximum correntropy criterion. It applied the loss function to the ELM model to improve its robustness in regression prediction. It proposed a method to estimate the percentage of noise in the actual sample, and gave the degree of noise pollution in the sample before established the load forecasting model. The simulation results show that the proposed algorithm model has better robustness and prediction accuracy when the outlier is more than 12%. [ABSTRACT FROM AUTHOR]

The recommendation algorithm based on graph neural network can extract the association relationship between users and goods. Traditional methods can’t extract this relationship. At present, most of these algorithms ignore the general prefe-rences in the review data of users and products. In order to solve this problem, this paper proposed a new method. This me-thod used the graph neural network to extract association relations, and the advantage of deep learning to extract the general preferences, and carried out feature fusion to improve the recommendation effect. This paper conducted comparative experiments and ablation experiments on four sets of public data sets to verify the effectiveness of the proposed method. The evaluation indexes include the recall rate and normalized discounted cumulative gain. Experiments show that this method is more effective than the existing algorithms. The feature fusion of the two networks can improve the recommendation effect. [ABSTRACT FROM AUTHOR]

This paper proposed an improved sparrow search algorithm（ISSA） to minimize makespan of Job-Shop scheduling problem（JSP）. Firstly, it formed the corresponding relationship between the discrete decision space of JSP and the continuous search space of sparrow search algorithm（SSA） by designing an effective code conversion mode. Then, since the SSA is easy to fall into local optimum, this paper improved SSA based on the idea of quantum computation, sine cosine search and decreasing vigilance strategy. Meanwhile, to make up for the lack of depth mining ability of SSA in solving discrete problems, it introduced multi-neighbor search and Gaussian disturbance strategy. Finally, this paper provided comparative experiments of 10 test questions of FT and LA series, 6 algorithms and 2 application examples. The results show that ISSA can obtain better mi-nimum value, average value and optimization success rate when solving JSP, which verifies the effectiveness of ISSA in solving JSP. [ABSTRACT FROM AUTHOR]

In order to ensure that the participants of the agricultural supply chain that joined the blockchain follow the agreement to share the initial true information, this paper built the agricultural supply chain framework of block chain technology to solve the problems of information asymmetry and lack of trust in the agricultural supply chain. It built an evolutionary game model for whether the participants of the agricultural supply chain adopting block chain technology kept their promises, and analyzed the factors that influenced the initial real information shared by each entity following the blockchain node. This paper used MATLAB for simulation, the results show that when the trust level, synergistic benefits, incentives and punishments of joining the blockchain increase, the risk coefficient becomes lower, the participants in the agricultural product supply chain are more inclined to abide by the blockchain agreement, these factors affect the strategic choice of participants under the blockchain node. This paper clarifies the feasibility of applying blockchain technology to the agricultural product supply chain, provides countermeasures for all participants, and improves the overall efficiency of the agricultural product supply chain. [ABSTRACT FROM AUTHOR]

Copyright of Application Research of Computers / Jisuanji Yingyong Yanjiu is the property of Application Research of Computers Edition and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In order to recognize complex interactive actions, this paper proposed a feature learning method based on depth information, and used a two-layer classification strategy to solve similar action recognition problems. The method analyzed the depth image action sequence from the frequency domain, extracted the frequency domain features, used VAE for spatial feature compression representation, established HMM to simulate time series changes and performed the first layer action recognition. In order to solve the problem of similar actions, this paper introduced three-dimensional joint point features for the second layer action recognition. Experimental results show that these two types of features can effectively represent the meaning of gestures on the SBU-Kinect action data set, the strategy is simple and effective and can get high recognition accuracy. [ABSTRACT FROM AUTHOR]

In order to solve the verifiable problem of outsourced computation in remote untrusted environment, this paper proposed an identity-based authorized function linearly homomorphic signature scheme on lattices by combining identity-based signature, homomorphic signature and functional signature. At first, this scheme verified the calculation results and the computing behavior of the cloud server at the same time, and avoided the shortcomings of using public key certificates. Secondly,this paper proved the proposed scheme to satisfy weakly unforgeability against adaptive chosen-identity, chosen-message and chosen-function attacks under the small integer solution assumption in the standard model. Finally, theory analysis shows that compared with the existing schemes, the proposed scheme is more efficient and safer. [ABSTRACT FROM AUTHOR]

In order to solve the problems of lack of state labels, lack of data samples and unbalanced distribution in equipment life prediction, this paper proposed an improved K-means algorithm based on PSO and a data optimization scheme based on traditional SMOTE. In the process of optimizing K-means algorithm, it combined the characteristics of particle swarm optimization algorithm, and improved the optimization efficiency of particle swarm optimization algorithm by giving the particle generation range of particle swarm optimization algorithm, so as to quickly judge the working state of the equipment. Then, by comparing the mean distance of samples in the same cluster with the distance from the sample to the center, it established an improved SMOTE algorithm, and avoided the calculation error caused by the imbalance of samples by adding the number of minority samples. Finally, this paper used AdaBoost integrated optimization KNN algorithm to improve the classification effect, and by fitting the equipment life curve, it could better predict the equipment health level and future life. The example shows that the proposed model can effectively predict the health status of equipment under small sample unbalanced data. [ABSTRACT FROM AUTHOR]

In order to solve the problem that the label semantic wasn’t fully utilized in cross modal hash retrieval method, which led to weak discrimination and low retrieval accuracy of hash code, this paper proposed a hash method named DICH-SSP (discriminant cross-modal hash with semantic similarity preservation). It projected the feature of heterogeneous modalities into a common subspace, and embedded the discriminant information and latent correlation of labels into the common subspace combined with multi-label kernel discriminant analysis. Then, it minimized the quantization error between the common subspace and the hash code to improve the discrimination of the hash code. In addition, this paper constructed the semantic similarity matrix by calculating the similarity between labels. DICH-SSP preserved the semantic similarity into hash codes, which could further improve the retrieval accuracy of hash codes. Extensive experiments on the three benchmark data sets of LabelMe,MIRFlickr-25 k, and NUS-WIDE demonstrate the effectiveness of DICH-SSP. [ABSTRACT FROM AUTHOR]

At present, there are few researches on aspect level sentiment analysis based on convolutional neural network and recurrent neural network, which take into account the syntactic structure of sentences and the syntax distance of words, and convolutional neural network and recurrent neural network can not effectively deal with the data of graph structure. To solve these problems, this paper proposed an aspect level sentiment classification model based on distance and graph convolution network. Firstly, this paper designed a two-layer bi-directional long short-term memory network with residual connection for the model to extract the context information of sentences. Then, the model obtained the weight of syntax distance of words and constructed the adjacency matrix from the syntactic dependency tree. Finally, the model used the graph convolution network to extract the sentiment features from context information, syntax distance weight and adjacency matrix. The experimental results show that the model is effective and can get better performance. [ABSTRACT FROM AUTHOR]

Supervised super-resolution algorithms require plentiful paired-image for networks training, resulting in poor visual recovery effects when processing real-world low-resolution images. To solve this problem, this paper proposed an improved semi supervised algorithm based on CycleGAN, donated as Cycle-SRNet. Firstly the degradation model generated images similar to real world low-resolution faces for network training, and then the reconstruction model restored high resolution face image. Furthermore, this paper introduced perceptual loss function to ensure the structural similarity of the faces so as to recover the facial features better. Experimental results show that the algorithm can restore fuzzy low-resolution video surveillance face images into recognizable face images, and it exceeds SRCNN, SRGAN and CinCGAN in terms of FID, PSNR and SSIM indicators. [ABSTRACT FROM AUTHOR]

In order to improve the accuracy and quality of the personalized recommendation system, this paper constructed a collaborative filtering recommendation algorithm based on SVD and intuitionistic fuzzy clustering ( SVDIFC-CF) to address the information overload and data sparsity problems in traditional recommendation algorithms. The algorithm firstly introduced SVD to fill in the original matrix after dimensionality reduction. Then it used the user-item preference matrix to perform intuitionistic fuzzy clustering of users. Finally it calculated the top N users with the highest similarity to the target user, and found the items that users were most interested in as recommendation results. This paper used the Movie Lens and Jester datasets to verify the effectiveness of the algorithm. The experimental results show that compared with the traditional recommendation algorithm, this algorithm can effectively solve the problem of data sparseness and cold start, and improve the accuracy and quality of recommendation. [ABSTRACT FROM AUTHOR]

Disaster preparedness measures can prepare for disaster relief to ensure that emergency supplies can reach the disaster area in a timely and effective manner. This paper proposed a bi-level programming emergency resource scheduling location-routing model with considering preparedness ( BLERSLR-P) In the model, the upper level aimed to minimize the total cost of supply station construction and operation, while the lower level aimed to minimize the cost of distribution route. This paper designed an improved bi-level salp swarm genetic algorithm (BLSSAGA) to solve this problem, combined the concept of iterative division to update the leader position, used the natural exponential inertia weight strategy to modify the control factor, used the chaotic map to update the follower position, and used Taguchi analysis to obtain reasonable values of parameters. Finally, this paper used bi-level salp swarm genetic algorithm ( BLSSAGA) and genetic particle swarm optimization algorithm ( GAPSO), particle swarm optimization algorithm( PSO), and immune optimization algorithm(IA ) to solve and compare and analyze the LRP data set in OR-Library, which verified the feasibility and effectiveness of the proposed model and algorithm. [ABSTRACT FROM AUTHOR]

Copyright of Application Research of Computers / Jisuanji Yingyong Yanjiu is the property of Application Research of Computers Edition and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In order to solve the problem that the siamese neural network is sensitive to position, complex background and intra-class differences due to the use of image-level feature metrics, this paper proposed a Dual Metric Siamese Neural Network (DM-SiameseNet) . DM-SiameseNet used image-level features and local features (local descriptors) to jointly represent each image, then learned feature maps based on two different levels of similarity measures, and finally used an adaptive fusion strategy to adaptively integrate two different measurement result represented by the level feature. Experimental results showed that the accuracy of the improved model was increased by 5.04% and 9.66%respectively, and was higher than the measurement methods that only use image-level feature representation or local descriptor representation on miniImageNet, TieredImageNet, Stanford Dogs, Stanford Cars and CUB-200 datasets. The experimental results prove that the proposed model not only considers the global features of the image, but also captures more effective local feature information in the image, which improves the generalization ability of the model. [ABSTRACT FROM AUTHOR]

This paper proposed an improved RRT（Rapidly-Explorative Random Tree） algorithm to solve the problem that the traditional target offset RRT was easy to fall into local optimum in the environment where there were many obstacles such as yarn racks, placing platforms and other packages during the package handling process. The collision detection model was established by geometric envelope method, and the distance coefficient g was added to keep a safe distance between the end of the manipulator and the obstacle; A random action selection strategy was proposed to select the expansion action before each expansion of the tree; A weakening target offset strategy was proposed to offset the extension direction according to the position of Xgoal relative to the Xnear and Xrand line; According to the vertical distance limit method and cubic spline interpolation, the initial path was optimized, and finally a smooth path was obtained. Through simulation comparison with RRT algorithm, M-RRT algorithm, RRT algorithm with added gravity coefficient and Improved RRT * algorithm, the maximum decrease in path length was 23.3%, the maximum decrease in time was 82.5%, and the distance from obstacles was always more than 50 mm. The results show that the algorithm improves the efficiency and safety of the manipulator handling. [ABSTRACT FROM AUTHOR]

In order to solve the problem that the existing cross-modal methods of chest X-ray images and diagnostic reports focus on global information alignment, which ignore the fine-grained semantic association between chest X-ray images and diagnostic reports, resulting in low retrieval accuracy and poor matching degree, this paper proposed a method named CDTCR (chest X-ray images and diagnostic reports for twin-towers cross-modal retrieval) Specifically, for fine-grained semantic representation, it proposed an image encoder composed of residual network to learn the fine-grained features of the image and a BERT model composed of Transformer to learn the fine-grained semantic features of the diagnostic report. In order to solve the problem of fine-grained semantic association, it also designed an information alignment strategy of two different granularity modes for the global image to sentence and local region to phrase, which solved the problem of insufficient fine-grained semantic association between different modes. The experimental results on a large-scale medical dataset MIMIC-CXR show that CDTCR has higher retrieval accuracy and better interpretation than the existing cross-modal retrieval methods. [ABSTRACT FROM AUTHOR]

To solve the problem of combustion state recognition in solid oxide fuel cell (SOFC) combustor, this paper proposed a combustion state recognition method based on attention mechanism and image feature pyramid. The method adopted the adaptive gamma correction with weighting distribution (AGCWD) to standardize the input images. It combined two single full connections with 1×1 convolution to replace the squeeze and excitation structure, and proposed a hybrid attention structure combined with spatial attention structure to enhance the ability of feature extraction. To provide multi-scale information communication capability, it constructed the multi-scale bidirectional fusion pyramid by means of bidirectional computation and multi-scale fusion. The experimental results show that the proposed method reaches 99.22% accuracy under the premise of 3.98 M parameters and 397 M floating point operations (FLOPs), and effectively identifies the combustion state in SOFC combustor. [ABSTRACT FROM AUTHOR]

At present, Chinese named entity recognition model has errors in recognizing entities with nested structures, so it can not be recognized accurately. The method based on span can find nested entities, but when detecting text, it often gene-rates span without entities, and can not clearly define the span boundary, which increases the model burden. To solve this problem, this paper proposed a Chinese nested named entity recognition model based on vocabulary fusion and span boundary detection. The model used multi-word fusion method to enhance text features, and it merged multiple lexical information related to the characters in the target sentence in the designed injection module. Ant then it integrated into BERT to obtain more comprehensive context information, and provided better span representation. Secondly, it added a span boundary detection method to divide the span boundary by predicting the first and last characters of the span by perceptual classifier. Experiments on public data sets show that the model can effectively improve the recognition accuracy. [ABSTRACT FROM AUTHOR]

This paper proposed a matching wavelet depth model transfer learning algorithm to solve the problems of low efficiency of feature identification and limited number of samples, which exist in traditional methods for small vibration fault diagnosis. Firstly, using Morse continuous wavelets to capture small changes in a one-dimensional fault signal by matching and up-dimensioning the signal to obtain a visual enhanced feature image. Then, this algorithm effectively transfers the source domain model of the depth transfer network. This model has efficient image learning experience and can reduce the number of training samples in the target domain. Finally, optimize the parameters of the process for this algorithm based on limited data in model transfer. The algorithm has proven to be highly generalisable, allowing the detection and localisation of minute features in multiple operating conditions and effectively reducing the reliance on data, greatly improving the speed of computing and diagnostic accuracy. [ABSTRACT FROM AUTHOR]

In order to solve the problems of uneven resource utilization and long task completion time in complex task group scheduling, this paper constructed a complex task group resource scheduling model to minimize the mean square error of resource load and the task group completion time, and proposed a multi-objective optimization algorithm based on boundary range local search and NSGA-Ⅱ, called BRLSN. The algorithm used an effective coding method and cross mutation operator for iterative optimization, and constructed an elite retention strategy based on local search in boundary region to expand the search scope of the algorithm and preserved good individuals in the population. Experimental results show that compared with other multi-objective algorithms, the convergence and diversity of BRLSN are significantly improved. At the same time, the algorithm convergence speed is faster, the population quality is higher, and the result value of the final objective function is obviously optimized. [ABSTRACT FROM AUTHOR]

In order to solve the problem that the driver is difficult to monitor and input complex during the running process, this paper proposed and implemented a driver fuzzing tool DrgenFuzzer based on the fitness and input constraint model. The tool utilized the kernel tracing technology combined with the static analysis of binary programs to realize the information monitoring of the driver operation, analyzed the driver interface parameters, designed a sample constraint scheme, proposed a new fitness calculation scheme and a crossover mutation scheme. Experiments show that compared with the commonly used kernel fuzzing tools, the test success rate of samples generated by this tool after the input constraint model is more than 10 times that of other tools, and the quality of the generated samples is higher. This tool performed a fuzzy test on the driver and exploits the null pointer reference vulnerability in the i2c driver. DrgenFuzzer can effectively guide and standardize sample generation, improve the success rate and operating efficiency of sample testing, and enhance the vulnerability mining ability. [ABSTRACT FROM AUTHOR]

Most existing traditional congestion control algorithms are difficult to adapt to the highly dynamic link environment of wireless Ad hoc network. In order to solve the above problem, this paper proposed a method of improving the performance of congestion control based on deep reinforcement learning, Enhanced-CC. It conducted a preliminary detection of the congestion window by using the traditional congestion control algorithm. On this basis, the method used deep reinforcement technology to learn the real-time optimal congestion window range of the link. When the congestion window calculated by the traditional congestion control algorithm was too large or too small, the method adjusted the congestion window, so that the sending rate could match the highly dynamic link bandwidth, and the method could improve the transmission performance of the traditional congestion control algorithm. The experimental results show that Enhanced-CC can significantly improve the performance of traditional congestion control algorithms such as BBR, CUBIC, Westwood, Reno, and is superior to the performance of fully learning based congestion control algorithms such as PCC and PCC Vivace, and the combination of deep reinforcement learning and traditional congestion control algorithms such as Orca and DeepCC. [ABSTRACT FROM AUTHOR]

Large-scale spiking neural network (SNN) parallel simulation is an important means to explore brain function. The difficulty lies in properly mapping the load to the multi-process parallel computing platform and improving the simulation speed. In order to solve this problem, this paper proposed a SNN load balancing method based on united weight hypergraph partitioning, which could solve the problem of balancing computing load and communication load between processes in parallel computing, and improve the speed of SNN simulation. And it used sparse communication instead of collective communication to solve the problem of data redundancy in the process of event communication and improve communication efficiency. The experimental results show that this method makes the simulation time of the cortical microcircuit model with STDP synapse scale of 20% shortened by about 64.5% compared with the standard cyclic allocation algorithm, and about 57.4% compared with the ordinary supermap allocation algorithm. At the same time, the communication time is reduced by more than 90%. [ABSTRACT FROM AUTHOR]

In order to fix the integer ambiguity quickly and accurately, this paper proposed a new ambiguity search algorithm based on measurement accuracy constraints to solve the problems of the LAMBDA algorithm, such as the wide search range of the integer ambiguity and the low search efficiency. When the least squares problem got optimally weighted, the algorithm took the carrier phase measurement accuracy as the constraint condition to test the fixed ambiguity solution on the basis of the stan-dard differential evolution （DE） algorithm. This algorithm solved the influence of different satellite altitude angles, and could achieve 99% success rate when solving 3D integer ambiguity. Compared with the MLAMBDA algorithm, the standard DE algorithm and the adaptive weighted differential evolution （AWDE） algorithm, the proposed algorithm further improves the efficiency and success rate of ambiguity resolution. [ABSTRACT FROM AUTHOR]

In the consistency detection of business process discovery research, existing methods of multi-perspective alignment between event logs and process models can only obtain the optimal alignment of one trace with the process model at a time. Meanwhile the computation of heuristic function in obtaining the optimal alignment is complex, leading to low computation efficiency. To solve above problems, the paper proposed a multi-perspective alignment between batch traces of event log and process model based on trace minimum edit distance. Firstly, the study selected multiple traces in the event log to form batch traces, and used a process mining algorithm to obtain the log model of the batch traces. Then it obtained the product model of log model and process model and their transition system, which was the search space of batch traces. Then it designed a heuristic function based on the minimum edit distance between Petri net transition’s sequence set and the remaining traces to speed up the A* algorithm. Finally, the study designed a multi-perspective cost function that could adjust the weight of data and resource perspectives, and proposed a method of multi-perspective optimal alignment between each trace in the batch traces and the process model with the transition system of the product model. Compared with existing work, the simulation results show the proposed method takes up less memory space and uses less running time. This method improves the computation speed of heuristic function for optimal alignment, and obtains all optimal alignments of batch traces at one time, thus improves the efficiency of multi-perspective alignment between event logs and process models. [ABSTRACT FROM AUTHOR]

In order to meet customers’ individual needs for delivery methods and specific time in urban distribution, this paper introduced the concept of customer preference to describe the delivery needs of home delivery and self-pickup services, and established vehicles path optimization model with different delivery options with the optimization goal of minimizing the total ope-rating cost. Considering the complexity of the model, it introduced a variety of operators, and designed an adaptive large neighborhood search algorithm and improved to solve the problem. Finally, combined with a case analysis of a distribution case in Nan’an district of Chongqing city, the validity of the model and algorithm is verified. [ABSTRACT FROM AUTHOR]