International Journal of Information and Communication Technology (61 papers in press)
Fault Diagnosis Model Based on Adaptive Generalized Morphological Filtering and LLTSA-ELM
by Jie Xiao, Jingtao Li
Abstract: It is difficult for single feature to contain all the information needed to describe the running state of the equipment, while multi-features can contain more information about running state, but the redundancy between high-dimension features can easily reduce the accuracy of the classifier. Aimed at that,a fault diagnosis method for rolling bearings combining adaptive generalized morphological filter, Linear Local Tangent Space Alignment and Extreme Learning Machine (LLTSA-ELM) is proposed. Firstly, the rolling bearing vibration signals are filtered by adaptive generalized morphological filter. Secondly, the multi-domain features are extracted from filtered signal to construct high-dimensional features set of bearing.Thirdly, the dimension of high-dimensional features is reduced by maximum likelihood estimation (MLE) and LLTSA.Finally, the bearing condition monitoring model based on ELM is constructed by the reduced dimension features and then use it to analyze and diagnose the running state of bearing.
Keywords: Adaptive Generalized Morphological filter; LLTSA; Dimension Reduction; ELM; Fault Diagnosis.
Site-selection method of agricultural products logistics distribution center based on blockchain
by Xingui Liu, Ming Luo
Abstract: In order to overcome the problems of low on time delivery rate and high distribution cost existing in the existing location methods of agricultural products logistics distribution center, this paper proposes a new location method of agricultural products logistics distribution center based on blockchain. Based on the analysis of the basic problems affecting the location of agricultural products logistics distribution center, combined with the blockchain technology, the location model of agricultural products logistics distribution center based on input-output ratio was constructed. Combining the idea of mountain climbing algorithm and particle swarm optimization algorithm, the hybrid particle algorithm is used to solve the location model of agricultural products logistics distribution center, and the optimal location scheme is obtained. The experimental results show that the proposed method can effectively improve the on-time delivery rate and customer satisfaction, and reduce the logistics distribution cost. The maximum on time delivery rate is 97.4%.
Keywords: Blockchain; agricultural products; logistics distribution center; site-selection method; hybrid particle swarm optimization algorithm.
Research on self-driving tour path planning method based on Collaborative edge computing
by Zhongbin Wang
Abstract: In order to overcome the problem of inaccurate results of traditional self driving path planning methods, a new self driving path planning method based on Collaborative edge computing is designed and proposed. In this method, collaborative edge computing method is used to remove the abnormal data and improve the accuracy of path planning. From the point of view of optimizing network performance, the objective function of network channel capacity of candidate path and the multi-objective optimization model of path selection problem are established. Finally, the Nash negotiation axiom in game theory is combined to solve the multi-objective optimization model to realize the self driving travel path planning. Experimental results show that the proposed method can effectively remove abnormal data, the highest removal rate is 99.74%, and the planning efficiency is above 96%.
Keywords: collaborative edge computing; path planning method of self-driving tour; abnormal data; mapping relationship; objective function.
Nonlinear autoregressive neural network with exogenous input for an energy efficient non-cooperative target tracking in wireless sensor network
by Munjani Jayesh, Maulin Joshi
Abstract: The prediction algorithms have been studied as a part of target tracking applications for many years. The prediction algorithm helps to select appropriate nodes to achieve precise target locations while tracking. The only group of sensor nodes nearer the predicted location is activated to save network energy. The inaccurate prediction algorithm may hamper energy consumption by activating inappropriate nodes resulting in a target loss. We propose a nonlinear autoregressive neural network with exogenous input (NARX)-based target-tracking algorithm that improves tracking accuracy
and energy efficiency. The proposed algorithm uses vehicle location time series and exogenous vehicle velocity time series as inputs and exerts accurate prediction location for given non-cooperative manoeuvring targets. The proposed algorithm is evaluated in terms of average prediction error, total network energy used, and the count of a target loss with state of art. The experiment outcome proves that the proposed novel NARX-based tracking algorithm outperforms and saves up to 26% of network energy with up to 83% reduction in tracking error compared to existing target tracking algorithms.
Keywords: wireless sensor network; WSN; non-cooperative target tracking; energy-efficient target tracking; prediction algorithm; sensor node selection; nonlinear autoregressive neural network.
Finding and identifying expert team members in open source environments
by Hani Bani-Salameh, Muntaha Bnyan, Fatima Abu Hjeela
Abstract: Team members working in open-source development (OSD) environments, often are geographically distributed developing their software projects. For any software project to succeed, it is important to find the right experts who possess all the required skills and expertise. Searching for help and locating the right experts for each task are essential as well as extremely complicated tasks in OS environments. This article explores the problem of identifying experts in open-source environments. It focuses on mining developers mailing lists to understand the developers social structure and point out the experts involved in the development process. We identify the experts using the interaction factor (number of interactions), which means that the higher the number of interactions, the more active the team member. Results show that central members can be paramount experts
in the network. Also, it shows that there are groups that can be defined by similarity based on members connections. It concluded that the structure and location methods are very useful approaches in experts prediction, and the proposed methodology can serve as an effective approach that OS project coordinators may use to identify experts.
Keywords: interactions; experts; teams; centrality; degree; betweenness; closeness; open-source; OS; environment; knowledge.
Online distance music teaching platform based on internet plus
by Wanshu Luo
Abstract: In order to overcome the low response speed and security problems of traditional music teaching platform, a new online distance music teaching platform based on internet plus is proposed. Based on internet plus, online distance music teaching server and online distance music teaching client is designed, to complete the hardware design of the platform. Through the construction of students personalised learning evaluation model and online distance music teaching program, the software design of the platform is completed. Combined with the platforms hardware and software, the design of online distance music teaching platform is realised. The experimental results show that online music teaching platform based on internet plus can shorten response time, and the response time is only 290 ms, which greatly improves the security of the platform.
Keywords: internet plus; online education; music teaching platform; evaluation model.
Traffic flow prediction model of urban traffic congestion period based on internet of vehicles technology
by Xiaofeng Shi, Yaohong Zhao
Abstract: There are some problems in the existing traffic flow forecasting models, such as low prediction accuracy and high time cost. RFID technology is used to transmit traffic flow data information during urban traffic congestion, and extract information to control the running state of vehicles on the road. The basic parameters of traffic flow are set, vehicle RFID data source is used to preprocess duplicate data and missing data of traffic flow, and differential stationarity and normalisation are processed; LSTM neural network is used to train traffic flow data iteratively and output estimate results. The comparison shows that the MAPE, RMSE and Mae of the proposed model are 12.34%, 23.18% and 15.87% respectively, which improves the prediction accuracy and the shortest prediction time is about 22 ms.
Keywords: internet of vehicles technology; traffic flow; radio frequency identification technology; LSTM neural network; normalisation processing.
Research on a feature fusion-based image recognition algorithm for facial expression
by Yilihamu Yaermaimaiti, Tusongjiang Kari
Abstract: In order to solve the problem that recognition rate of facial expression images is easily affected by non-uniform illumination factors, an improved face recognition algorithm is proposed in this paper. Firstly, a facial expression image with Log-Gabor feature vectors of multiple scales and directions is extracted from a face image, and then all the Log-Gabor feature vectors are blocked in a unified way. Secondly, gist algorithm is applied to extract gist feature blocks from Log-Gabor feature vector image, and then all those blocks are cascaded together as the feature vectors of a face expression sample. The fused feature vectors of the face expression sample are trained as the input feature of the stacked auto-encoder (SAE). Finally, the trained expression features are input into the classifier for recognition to obtain the final recognition result. Whether it is in the facial expression database JAFFE or the Uyghur facial expression database, its facial expression recognition rate is the highest, which verifies the superiority of the algorithm we put out in this paper.
Keywords: feature vectors; feature blocks; stacked auto-encoder; SAE; Uyghur facial expression database; UFED.
Research on wireless sensor privacy data measurement and classification model based on IoT technology
by Jianye Wang, Chunsheng Zhuang, Xin Liu, Shunli Wu, Ding Wang, Haoping An
Abstract: Traditional wireless sensor privacy data measurement and classification process has the problems of low accuracy and long time-consuming. A new wireless sensor privacy data measurement and classification model is proposed. The measurement module of the model establishes two levels of privacy measurement elements from three dimensions, calculates the weight of level privacy elements by using Shannon information entropy, and obtains the privacy quantity of each data privacy measurement element in the data set; With the help of Internet of things technology, the privacy data measurement and classification model of wireless sensor is designed. The experimental results show that the misjudgement rate of this model is very low, less than 5%, and the misjudgement error is kept below 10%. Using this model to protect the wireless sensor privacy data, the data running cost and time cost show a low level, has a certain application value.
Keywords: wireless sensor; privacy data; privacy elements; data measurement; data classification.
Visual segmentation of diagnosis image of pulmonary nodules with vascular adhesion based on convolution neural network
by Yingying Zhao, Chunxia Zhao, XueKun Song
Abstract: Because of the low grey value of the background area in the diagnosis image of pulmonary nodules with vascular adhesions, the traditional visual segmentation method is weak for image feature recognition, which results in the unsatisfactory visual segmentation effect. A visual segmentation method based on convolution neural network is proposed for the diagnosis image of pulmonary nodules with vascular adhesions. Three modes are added to the original convolution neural network through the filter, and the convolution neural network is used to fuse the diagnosis image of pulmonary nodules with duct adhesion. The fusion results were processed by the fuzzy c-means method to complete the visual segmentation of the diagnosis image of
pulmonary nodules with vascular adhesion. The simulation results show that the proposed method can quickly and accurately complete the visual segmentation of the diagnosis image of pulmonary nodules with vascular adhesion, and has strong adaptability.
Keywords: convolution neural network; pulmonary nodules with vascular adhesion; visual segmentation of diagnosis image.
A collection method of motion video motion track based on fuzzy clustering algorithm
by Honglan Yang, Guan He
Abstract: In order to overcome the problems of low spatial accuracy, poor noise reduction ability and high spatial distortion rate of traditional methods, a motion video motion trajectory collection method based on fuzzy clustering algorithm is proposed. The fuzzy clustering algorithm is used to pre-process the motion video to improve the compression ratio of the motion video. The motion point accumulation method is used to extract the image sequences in the motion video, the motion targets are extracted, and the trajectory points of the motion targets are labeled. Establish a new motion trajectory range, and realise the collection of the motion trajectory within the marked range. The experimental results shows that the minimum noise value of the proposed method is only 0.02dB and 0.03dB, the spatial distortion rate of designed method is lower, it has better anti-noise effect, and more accurate motion
trajectory acquisition results.
Keywords: moving target extraction; track point annotation; motion trajectory; descriptor; integral invariant.
Facial expression recognition of aerobics athletes based on CNN and HOG dual channel feature fusion
by Shitao Wang, Jing Li
Abstract: The problem of low feature extraction accuracy and low recognition accuracy in facial expression recognition of aerobics athletes is presented. Propose a recognition method for fusion CNN and HOG dual channel features. The basic principle of the HOG is analysed, and the facial expression image of aerobics athletes is processed by grey level with the help of local binary mode. The pixel gradient intensity value in each small image is obtained, and all the intensity values are fused. Lagrange formula is used to transform high-dimensional features, support vector machine is used to classify facial expression images, and feature points are used as CNN, to process feature points according to network input and regularisation regression is used to realise facial expression recognition of aerobics athletes. The experimental results show that the accuracy of feature extraction is 97% and the recognition accuracy is always higher than 90%.
Keywords: CNN; HOG; local binary mode; pixel gradient intensity value; facial expression.
Feature extraction method of football fouls based on deep learning algorithm
by Weicheng Ma, Yanfei Lv
Abstract: In order to overcome the problems of abnormal detection and low accuracy in the process of football foul feature extraction, this paper proposes a football foul feature extraction method based on deep learning algorithm to accurately identify the fouls in the process of normal competition. In this method, the background is eliminated by the difference between the input image and the background image, so as to obtain the effective detection target. According to the characteristics of football competition, the human motion tracking algorithm is proposed. Through the template representation, candidate target representation, similarity measurement calculation and search strategy, the dynamic target is tracked in real-time, and its dynamic information is obtained. Finally, the star skeleton feature is used to extract the football foul action feature, and the image feature is transformed into available data to realise the data extraction of action feature. The experimental results show that the proposed method can detect the target with low accuracy.
Keywords: deep learning; human motion; action recognition; mean shift algorithm; background subtraction.
Research on dynamic parameter identification method of shallow reservoir based on kalman filter
by Shaowei Zhang, Rongwang Yin
Abstract: The identification method of reservoir parameters has the problems of low recognition accuracy and timeliness. A dynamic parameter identification method of shallow reservoir based on kalman filter is proposed. The history fitting method is used to establish and adjust the shallow reservoir model, and the parameters and range of the reservoir model are continuously adjusted according to the actual observation data of the shallow reservoir. Kalman filter is used to filter the data of shallow reservoir to filter out the noise and interference information. Then the dynamic parameters of shallow reservoir are identified by the method of water resistivity shale content discrimination, and the state of shallow reservoir is reflected by the shallow water resistivity. The comparison shows that the average recognition accuracy of the method can reach 95.2%, the recognition process takes only 22 seconds at most, and its recall precision value level is always high.
Keywords: historical fitting; least squares objective function; reservoir model; kalman filtering; parameter identification.
Research on reliability analysis of catenary model based on the fusion particle swarm least square support vector machine algorithm
by Haigang Zhang, Xuan Chen, Piao Liu, Decheng Zhao, Bulai Wang, Jinbai Zou, Minglai Yang
Abstract: The reliability of contact network is always an important part of the reliability analysis of traction power supply system. In this paper, combined with the failure rate data of the main parts of the contact network, the small sample data is expanded by Bootstrap non-parametric regeneration sampling method using the Fused Particle Swarm Least Squares Support Vector Machine (PSOLSSVM) algorithm to provide training set data for particle swarm optimisation. Mann and Schuer and Fertig fit and test the two-parameter Weibull distribution of the model based on MATLAB. Select the key components such as load cable and insulator, and combine the data to establish a model to characterise the overall reliability of the contact network. Based on the established model, the relevant parameters of each main part are estimated separately, which accords with the actual situation.
Keywords: Weibull distribution; least squares support vector machine; parameter fitting; bootstrap.
Intelligent traffic assignment method of urban traffic network based on deep reinforcement learning
by Zhiyong Jing, Huanlong Zhang
Abstract: In order to overcome the problems of high relative error and long response time of traditional methods, a new intelligent traffic flow allocation method based on deep reinforcement learning is proposed. In this method, deep reinforcement learning is introduced, and experience pool technology is used to obtain and retain samples in a certain stage to train urban traffic network. The complete track is divided into several independent state action pairs, and the sample database is established. In a certain range, the vehicle congestion density is simplified to the degree of congestion. When the starting point and the end point are known, all traffic demands between the two points are calculated allocation, intelligent traffic network traffic assignment is realised. Experimental results show that the average relative error of passenger travel time is 12.34%, the traffic flow prediction indexes are the lowest, the allocation time is the highest, which is 0.878 s.
Keywords: Deep reinforcement learning; urban traffic; network traffic; intelligent distribution.
Potato late blight disease detection using convolutional neural network
by Mominul Islam, Md. Ashraful Islam, Ahsan Habib
Abstract: This paper proposes a convolutional neural network-based deep learning model to classify and detect the infectious potato leaves suffering from late blight disease. The proposed model has two classifiers - the potato leaf classifier and the late blight disease classifier. Both healthy and diseased plant leaf images taken from the plantVillage dataset and real-time images are used to train, validate and test the classifiers. A total of 4,680 and 1,470 plant leaf images are used for the two classifiers, respectively. The potato leaf classification accuracy of the proposed model is 97.12%. The proposed CNN model also provides an accuracy of 98.62% while identifying late blight disease. The ten-fold cross-validation technique is used to observe the performance of the proposed late blight classifier and then compared with other cutting-edge approaches. In observation, it has been shown that the proposed technique outperformed many other existing techniques.
Keywords: late blight; convolutional neural network; deep learning; image processing; image augmentation.
Classification of existing mobile cross-platform approaches and proposal of decision support criteria
by Ayoub Korchi, Mohamed Karim Khachouch, Younes Lakhrissi, Nisrine El Marzouki, Aniss Moumen, Mohammed El Mohajir
Abstract: The smartphone market has known an exponential growth since 2007, with the apparition of the first Apple phone. Nowadays, developing an application that targets all existing mobile platforms, becomes a tedious task for developers, due to the diversity of mobile platforms, their tools. For that, cross-platform approach with its various sub-approaches has shown its strength in reducing projects cost and time respecting the slogan develop once and deploy everywhere. This paper aims to compare the mobile app developments approaches and suggests a decisional framework to choose the adequate one to get an application respecting the clients need with a low cost and time. This frameworks criteria have a huge impact on the eventual cost, time, and success of an application building. If developers fail to match an app demands to the right development approach, it can turn their project into a certain failure.
Keywords: OS; development; cross-platform approaches; MDA.
Converged communication method of multi-source data about underground equipment based on Internet of things
by Shunli Wu, Haoping An, Yan Gao, Jianye Wang, Zhan Su
Abstract: The difference and incompatibility of multi-source data in complex environment lead to high bit error rate and low efficiency in communication of downhole equipment. Therefore, a multi-source data fusion communication method for underground equipment based on internet of things is proposed. Using multi-dimensional information acquisition system, the feature recognition model of multi-source data of underground equipment is established. According to the distribution of boundary mesh objects among clusters, the boundary object set of data distribution is obtained, and the data features are extracted. The balance degree of multi-source data is calculated, and the data is represented in grid form to realise fusion. The communication process is optimised according to the convergence result. Experimental results show that the proposed method has high efficiency (the highest efficiency is 97%) and low bit error rate (the lowest bit error rate is 0), which improves the efficiency of multi-source data fusion of downhole equipment.
Keywords: internet of things; underground equipment; multi-source data; convergence; channel configuration.
Information technologies of the Russian-Cuban GNSS service
by Ilia Bezrukov, Vladislav Yakovlev, Dmitry Marshalov, Yuri Bondarenko, Alexander Salnikov, Omar Pons Rodriguez
Abstract: We present the hardware and software implementation of the information technologies for the GNSS service of the Russian-Cuban co-located geodetic station. The service is equipped with geodetic and meteorological instruments with data acquisition and transmission systems and allows to conduct high-precision observations of the GPS and GLONASS global navigation satellite systems (GNSS) as well as meteorological measurements. The coordinates of the station are refined while processing GNSS observations and according to the change in coordinates, the movement of Cuban tectonic plates is estimated. Information technologies of the GNSS service is one of the key elements required for conducting regular automated geodetic and meteorological measurements. Information technologies is a group of computing and networking equipment that provides preliminary processing and transfer of observations to the data processing centre at the Institute of Applied Astronomy in St. Petersburg. It also allows the remote monitoring, control, and maintenance of GNSS service scientific instruments.
Keywords: remote control and monitoring; virtual private network; VPN; informational security; GNSS service; geodesy; applied geophysics.
A PRI estimation and signal deinterleaving method based on density-based clustering
by Lei Wang, Zhiyong Zhang, Tianyu Li, Tianhe Zhang
Abstract: In the existing statistics-based PRI estimation method, it is difficult to improve the PRI estimation accuracy due to the contradiction between the width of the statistical interval and the PRI extraction accuracy. In order to improve the accuracy of PRI estimation, a radar signal PRI estimation and deinterleaving method based on the density-based clustering is proposed in this paper. The dense area of the time of arrival (TOA) difference sequence near the true PRI value is extracted out by density-based clustering, take the intra-class mean value as the PRI estimation value and the intra-class point dispersion interval length as the PRI jitter amplitude. Combined with the sequence searching method with dynamic tolerance, the pulse sequence with a large number of pulses and small PRI jitter is preferentially extracted, which can improve the accuracy of signal deinterleaving. The simulation results show that the proposed method can significantly improve the accuracy of PRI estimation and the success rate of signal deinterleaving in the case of PRI jitter and false pulse interference.
Keywords: radar emitters; radar signals; pulse repetition interval; PRI; PRI estimation; signal deinterleaving; density-based clustering; DBSCAN; time of arrival; TOA; PRI jitter.
Intelligent recommendation method for personalised tourist attractions based on cloud computing technology
by Changchun Guan, Jinhua Luo
Abstract: In order to overcome the problems of poor recommendation results and long travel time of traditional personalised tourist attractions recommendation methods, this paper proposes an intelligent personalised tourist attractions recommendation method based on cloud computing technology. The method constructs user interest model based on knowledge map vectorisation and user interest vectorisation. In the algorithm recommendation module, based on Hadoop cloud platform, Maple-Duce is parallelised, and the Bayesian network is used to predict the users preference for items. The probability is presented to show the possibility of the users preference for items, and the user is recommended according to the probability, so as to complete the personalised intelligent recommendation design for tourist attractions. The experimental results show that the personalised tourist attractions intelligent recommendation method has the highest recommendation accuracy up to 99%, and reduces the travel time, the minimum time is 430 min, which is feasible to some extent.
Keywords: cloud computing technology; individualisation; tourist attractions; intelligent recommendation.
Research on operational effectiveness of air and missile defence in maritime stronghold based on queuing theory
by Wenfei Zhao, Kenan Teng, Jian Chen, Yan Wang
Abstract: For the complexity and uncertainty of the enemy situation faced by the defence decision-making in the maritime stronghold, this study adopts the queuing theory to study the process of air and missile defence operations in the maritime stronghold, quantitatively analyses the state transfer probability of the stochastic service system for air and missile defence operations in the maritime stronghold, constructs a mixed system queuing model of multiple service stations arriving in batches, and gives the evaluation model of operational effectiveness for air and missile defence. Finally, the validity of the model and algorithm is verified by arithmetic simulation.
Keywords: maritime stronghold; air and missile defence; queuing theory; operational effectiveness.
Group popular travel route recommendation method based on dynamic clustering
by Yanhua Guo
Abstract: This paper proposes a new group popular travel route recommendation method based on dynamic clustering. Based on the recommended pattern graph, the decreasing function of tourist interest and the score matrix of interest preference are calculated. And the dynamic clustering method is used to construct the dynamic mining model of group passenger preference data to obtain the tourist preference data. Based on the preference data, the hot areas of tourist routes are divided, and the Markov model is used to calculate the transfer probability of tourist routes, and the final result of tourist route recommendation is obtained. Experimental results show that, compared with traditional recommendation methods, the proposed method has higher recommendation accuracy and efficiency, and the highest recommendation accuracy and efficiency can reach 97% and 98%. Therefore, the proposed method is more effective.
Keywords: Dynamic clustering; popular group; tourist route; recommendation method.
Algorithm for interference filtering of Wi-Fi gesture recognition
by Wei Han, Jing Yu
Abstract: Nowadays, On the one hand, with the continuous improvement of computer technology, human-computer interaction becomes more and more important in people's life. Among them, gesture, as an intuitive human language, has become an important way of human-computer interaction. On the other hand, as an important branch of mobile network, WLAN has gradually developed into an irreplaceable technology in indoor communication. This paper based on the relevant knowledge of the wireless channel state information (CSI), put forward under the environment of Wi-Fi, with the help of fertility carrier amplitude information provided by the CSI for fine-grained gesture recognition. Because of interference, a gesture recognition system based on Wi-Fi accuracy and robustness to ascend, therefore this paper proposes a Wi-Fi gesture recognition interference Filter algorithm, using Butterworth low-pass Filter and principal component analysis (PCA), combining to the CSI of raw data denoising processing, filtering CSI noise in the raw data. On the basis of the recognition and machine learning the results verify the robustness and accuracy of the algorithm.
Keywords: channel state information; CSI; noise and interference filtering; gesture; recognition; machine learning.
Colossal pattern extraction using optimised length constraints based on differential evolutionary arithmetic optimisation algorithm
by T. Sreenivasula Reddy, R. Sathya, Mallikharjunarao Nuka
Abstract: Extracting large amounts of information and knowledge from a large database is a trivial task. Existing bulk item mining algorithms for an extensive database are systematic and mathematically expensive and cannot be used for large-scale mining with interruptions. In this paper, the problem of mining the colossal patterns (CPs) is solved by using optimised length constraints (LCs). First, we describe the minimum LC and maximum LC problems and connect them to the optimal LC by identifying the optimal threshold values. Here, the differential evolutionary arithmetic optimisation algorithm (DAOA) is used to find the optimal threshold values of the constraints and extract the colossal patterns. The effectiveness of the proposed algorithm is proven by various experiments using different biological datasets.
Keywords: biological databases; colossal patterns; differential evolutionary arithmetic optimisation algorithm; DAOA; massive data itemset; optimised length constraints.
Design of a new type of logistics handling robot based on STM32
by Xinliang Cheng, Jinyun Jiang, WanXin Fu, Shiyi Ying, Xiaoliang Jiang
Abstract: In order to improve the stability of the logistics handling robot and the accuracy of grasping logistics materials, this paper proposes a new type of logistics handling robot, which is designed by using the combination of PID closed-loop control algorithm and visual recognition technology. In the part of robot software, we use the PID closed-loop control algorithm to make real-time control of the robot, so that it can accurately patrol the line, and use the raspberry Pi call recognition algorithm (opencv) to complete the identification of two-dimensional code, bar code and logistics materials. In the structural part, we improved the five-degree-of-freedom articulated manipulator to realise three-dimensional grasping and ensure fast and accurate grasping and placing of logistics materials. In order to verify the performance of the robot, we design a mini logistics warehouse environment to simulate the real experimental scene. The experimental results show that the designed logistics handling robot can accurately locate and place logistics materials, and complete the handling task according to the task requirements.
Keywords: logistics handling robot; PID; recognition; articulated manipulator; STM32; image.
Assessment and insurance of cyber risks as tools for ensuring information security of an organisation (on the example of Russia)
by Dmitry R. Sergeev, Oksana N. Suslyakova, Gulnaz F. Galieva, Elena E. Kukina, Olga Yu. Frantsisko
Abstract: The purpose of the study is to reveal the specifics of cyber risks as a source of reducing the information security of organisation, as well as to develop a methodological approach to the formation of tools for managing cyber risks based on their assessment and insurance. The authors analysed the dynamics of the growth of cyber risks in relation to small, medium and large organisations, and also assessed the possible scale of damage from their occurrence in the global economy. The article offers a methodological approach to the analysis of risk factors that affect the amount of possible damage and the likelihood of cyber risk. The authors formed an algorithm for managing the organisations cyber risks based on their assessment. The analysis of advanced foreign experience allowed the authors to determine the directions of its adaptation in the process of modernisation and improvement of the mechanism of cyber risk insurance.
Keywords: cyber risk; cyber incident; digital economy; cyber risk insurance; information security.
Securing BYOD environment from social and mobility related threats: the case of Nigerian banking sector
by Lizzy Oluwatoyin Ofusori, Prabhakar Rontala Subramaniam
Abstract: Globally, bring your own device (BYOD) is gradually gaining popularity in work environments and businesses. The benefits of BYOD, which include increased productivity, flexibility, and efficiency, have necessitated all sectors to adopt this trend to maximise the benefits. The banking sector, in particular, has been at the forefront of the adoption of BYOD, and employees are now enjoying the benefits. However, despite BYOD benefits, security threats and privacy invasion have been a major concern for individuals and organisations. Ofusori et al. (2018) have classified these threats into technical, social and mobility threats. Thus, this paper investigates the influence of social and mobility threats as it relates to BYOD phenomenon in the banking sector. Data was collected from Nigerian bank employees, and the study found that there are overlapping threats due to the influence of social threats over mobility threats. This study addresses the overlap security threats.
Keywords: security vulnerabilities; security measures; security practices; bring your own device; BYOD; mobile devices.
Online education big data mining method based on association rules
by Na Zhang
Abstract: In order to solve the problems of slow mining speed, high noise and poor data correlation in the existing online education big data mining methods, an online education big data mining method based on association rules is designed. Firstly, the recursive distance of the big data centre is determined, and the online education big data is extracted according to the calculation of fuzzy membership. Secondly, the covariance matrix is used to remove the noise in online education big data and reduce the dimension. Finally, calculate the confidence and support of online education big data association rules, determine the association strength between online education big data in the set, and complete data mining. The experimental results show that the mining speed of this method is significantly improved, the longest time is no more than 4 s, and the data mining is highly correlated.
Keywords: association rules; online education; big data; Euclidean distance; uniformity.
Analysing the Algerian social movement through Twitter
by Meriem Laifa, Djamila Mohdeb, Mouhoub Belazzoug
Abstract: Technology has altered collective actions guidance resulting in a new regulatory frame for action. For the sake of being successful in a social movement, people plan and advertise in advance to encourage and gather greater participation to strengthen the influence of crowds. For this, social media offers exceptional opportunities to organise masses of people into actions with lower participation expenses, and to foster new repositories of information and actions that go beyond communities offline. While most contemporary social movements have been studied from different perspectives, the Algerian social movement (i.e., Hirak) was overlooked in the literature. This paper presents a distinctive foundation for understanding the Algerian Hirak through analysing Twitter data. The used approach is established mainly at the intersection of sociology and data analysis, with the intention to generate an improved discernment of this movement. Promising future research directions are also discussed in this paper.
Keywords: social movements; social media; Algerian Hirak; natural language processing; Twitter; Algerian Social Movement.
Received signal strength-based power map generation in a 2-D obstructed wireless sensor network
by Mrinmoy Sen, Indrajit Banerjee, Tuhina Samanta
Abstract: This paper analyses the effect of received signal strength (RSS) in efficient deployment, in presence of obstacles. We consider RSS based power values, so that x-y plane represents the spatial coordinates within a target field and z coordinates denote power values over the field. We plot the power values on the x-y coordinates, addressed as power map, having some peaks and falls: the peaks represent strong signals and the falls represent weak signals at the co-ordinates. It is intuitive that locations with strong signals are more suitable for communications. The falls in the power strength indicates that more sensor nodes are to be put for successful communications. We validate the proposed scheme via simulations as well as small-scale indoor and outdoor experiments with XBee sensor motes. We propose an algorithm to estimate the received power and analyse the estimated results with the results generated through the hardware test-bed.
Keywords: noisy channel; node deployment; power map; channel frequency; obstructed network.
Research on dynamic bidirectional security authentication of user identity in wireless sensor network based on improved des algorithm
by Xiaodong Mao, Haiyan Li, Shixin Sun
Abstract: In order to overcome the problems of time-consuming and weak attack defence of traditional authentication methods, this paper proposes a new dynamic bidirectional authentication method for user identity in wireless sensor network based on DES algorithm. This method uses DES algorithm to complete the user identity information encryption at the client-side through three steps: initial transposition, encryption operation and final transposition of identity information. In view of the server transmission to obtain the challenge value, combined with the one-time password DES to distinguish the authenticity of the server, the user identity authenticity verification is completed. Through the client-side and server-side legitimacy verification, the dynamic bidirectional security authentication of user identity is realised. The experimental results show that the proposed method is efficient, and the security of identity authentication is stronger, which provides theoretical support for the research of related fields.
Keywords: wireless sensor network; DES; user identity; bidirectional; authentication.
Which people are loyal followers of influencers? An exploratory study
by Javier A. Sánchez-Torres, Juan Sebastían Roldan-Gallego, Francisco-Javier Arroyo-Cañada, Ana María Argila-Irurita
Abstract: Influencers are tools implemented in digital marketing as a communication mechanism between the brand and its target; however, there are few studies that observe the relationship between the personality of the follower and their attitude towards the influencer. The objective of this study is to explore whether personality traits influence positive attitudes towards influencers. An empirical study was carried out in Spain and Colombia with a sample of 381 individuals and cause-effect relationships were analysed using the partial least squares methodology. The results show that extroversion and disordered personality traits are related to positive attitudes towards influencers and there could be some differences between genders, specifically men with a calm personality and women with a sympathetic personality
Keywords: influencers; personality; followers; social network analysis; internet marketing; digital marketing; partial least squares methodology; extroversion; disordered personality; calm personality; sympathetic personality.
An automatic correction system of singing intonation based on deep learning
by Hui Tang
Abstract: In order to solve the problems of low accuracy and slow correction speed in traditional singing intonation correction system, an automatic singing intonation correction system based on deep learning is proposed. In the hardware, floating-point DSP and TDSP-TF984 chip are selected as the core chips of automatic correction processor of singing intonation. The data input module and parameter calculation module of singing intonation are designed to improve the singing intonation data collector. In the software, the group delay estimation method is used to collect the singing intonation signal, and the deep learning algorithm is used to decompose the false component of the singing intonation signal, and the autocorrelation function and characteristic distribution operator of the singing intonation signal are obtained to realise the singing intonation signal correction. The experimental results show that the highest accuracy of the proposed system is about 97.8%, and the shortest correction time is about 1 s.
Keywords: deep learning; singing intonation; automatic correction; signal extraction; autocorrelation function.
A wireless sensor network node redeployment method based on improved leapfrog algorithm
by Bin Zhang, Xinhua Wang
Abstract: In order to overcome the problems of large errors and low average coverage of nodes in traditional node redeployment methods, a node redeployment method based on improved frog jump algorithm is designed in this paper. The output node path of wireless sensor network is determined by constructing the distribution node deployment model, then the leapfrog algorithm is improved by introducing virtual force algorithm, and the physical model of node deployment area is established, so as to optimise the node deployment process by using gravity and repulsion force, and redeploy static and dynamic nodes. Experiments show that the minimum node redeployment error of this method is only 0.01. When the energy of some nodes is exhausted, it still has relatively good coverage quality performance, and the average coverage rate reaches 75%, which proves that it not only ensures the network coverage quality, but also reduces the number of working nodes.
Keywords: improved leapfrog algorithm; wireless sensor network; node redeployment; node path.
Study on a method for capturing basketball player's layup motion based on grey level co-occurrence matrix
by Luojing Wang
Abstract: In the process of basketball players layup motion capture, the image blur leads to high layup motion capture error. Therefore, a basketball player layup motion capture method based on grey level co-occurrence matrix is proposed. The hue, saturation and brightness components of basketball players layup action image are unified greyed, and the fuzzy information in the image is transformed into different greyscale information; The Pearson correlation coefficient is used to analyse the correlation between each component after greying, and the grey information of fuzzy image is filtered by establishing grey co-occurrence matrix; By analysing the change of positioning coordinates of basketball players layup action in three-dimensional space, the core area of action capture is determined, and the key point position capture results are aggregated to realise the capture of basketball players layup action. The results show that the accuracy of the proposed method can reach 97.14%.
Keywords: grey level co-occurrence matrix; motion capture; grey processing; Pearson correlation coefficient; blur image; positioning coordinates.
Research on deduplication method of multiple relations based on hierarchical clustering algorithm
by Ying Wang, Weiwei Cheng, Chang Liu
Abstract: In order to overcome the problems of low efficiency and large error in traditional data deduplication methods, a multi relational data deduplication method based on hierarchical clustering algorithm is proposed. According to the inter class relationship information of duplicate data, different types of closely related class clusters are merged. Through hierarchical clustering algorithm, all the duplicate data are clustered according to the data similarity. After finding the similar class in the first level index, the super eigenvalue is used to complete the detection of multi relationship duplicate data. According to the specific situation at that time, the detected duplicate data is deleted by automatic, semi-automatic or manual methods. Experimental results show that the method has low error rate and good deletion effect, and improves the efficiency of multi relational data deduplication, with the highest deletion rate of 99%.
Keywords: hierarchical clustering; multi relational data repetition; super eigenvalue; inter class relationship.
Study on route selection method of scenic spot tourism based on GIS spatial model
by Yaqi Shi
Abstract: In order to overcome the problems of low coverage and high cost in traditional route selection methods for scenic spots, this paper proposes a new route selection method for scenic spots based on GIS spatial model. In the GIS spatial model, the weight scores of influencing factors are calculated. According to the weight calculation results, the weight values of various influencing factors are jointly superimposed on the grid of the GIS spatial model, the cumulative value of each grid is calculated, and then the weighted distance function of the scenic sightseeing route selection is established, to calculate the minimum distance from the grid to the scenic spot, so as to complete the route selection for scenic sightseeing. The experimental results show that the proposed method can improve the coverage of route selection and reduce the cost of route selection and the maximum coverage of route selection reaches 98%.
Keywords: GIS spatial model; sightseeing and tourism; route selection method; weighted distance function.
An efficient retrieval of relational database information based on knowledge graph
by Hongqin Zhu
Abstract: In order to optimise the effect of database information retrieval and overcome the problem of high index cost of traditional retrieval methods, an efficient retrieval design method of relational database based on knowledge graph is designed in this paper. Firstly, the distribution model of database information retrieval nodes is constructed to analyse the data structure, and the semantic characteristics of stored data of retrieval nodes are fused. Then the knowledge graph model analysis method is used to retrieve the database node attribute clustering and adaptive scheduling to realise the design of efficient information retrieval model for the database. The experimental results show that the index distance cost of this method is always below 5 x 1,010 byte, and the index distance cost is always below 40 s, which has certain application value.
Keywords: relational database; knowledge graph; indexing structure; semantic features; data graph; key words; information retrieval.
A deep mining method of enterprise financial risk data based on improved support vector machine
by Hui Sun, Li Fang
Abstract: Aiming at the problems of low accuracy of enterprise financial risk data mining, large amount of redundant data and low accuracy of data mining in traditional methods, an enterprise financial risk data depth mining method based on improved support vector machine is proposed. Design the deep mining route of financial risk data, fuzzy transform the financial risk data, and complete the preprocessing of enterprise financial risk data. According to the transformation results, by improving the linear separability and nonlinear separability of data in support vector machine, based on the improved support vector machine, the optimal hyperplane in these two cases is obtained to realise the classification of enterprise financial risk data. On this basis, RFID module is designed to complete the in-depth mining of enterprise financial risk data. The experimental results show that the data mining accuracy of the proposed method is 88%.
Keywords: improved support vector machine; financial risk; data mining; linearly separable; linearly inseparable.
A secure and integrated ontology-based fusion using multi-agent system
by Tarek Salah Sobh
Abstract: This study aims to handle ontology-based fusion and use multi-agent systems to obtain information fusion from multiple sources/sensors in a secure and integrated manner. Therefore, our objective is to produce a secure and integrated ontology-based fusion framework by using multi-agent. The agent system gets different props from using ontologies such as interoperability, reusability, and support. Here, fusion levels vary from the signal level that is low to the high knowledge level. Securing a multi-agent platform was introduced through a security system called 'SMASP'. The performance results show that the framework is almost idle while the user is composing the query. The workload is low on CPU and memory. This framework receives multiple data sources through cloudlet. Ontologies support a secure multi-agent system with different operations such as reasoner agents and query agents. Using the cloudlet architecture gives the flexibility to overcome intensive computing and sensitivity to latency.
Keywords: information fusion; multiple data sources; integrated framework; ontology; reasoning; cloudlet; agent security.
An efficient single unit for virtual-machine placement in cloud data centres
by Salam Ismaeel, Ali Miri, Ayman Al-Khazraji
Abstract: There are numerous energy minimisation plans are adopted in todays data centres (DCs). The highest important ones are those that depend on switching off unused physical machines (PMs). This is usually done by optimal distribution and/or reallocating of virtual machines (VMs) on the selected servers. While maintaining the quality of service (QoS) to ensure the performance of a DC. In this work, a novel server machine condition index (MCI) has been proposed, which includes all resources related to servers available in the DC using a single unit. The MCI represents a dynamic tool to compare services, increase effectiveness, reflect PM adequation, and ensure the optimal management of heterogeneous DC resources. The MCI will be used to convert the multi-objective VM allocation optimisation problem into a single-objective problem. This work will identify the MCI components and the way that can be used as a cloud resource unit, and modified VMP algorithms.
Keywords: power consumption; virtual-machine placement; cloud data centre; closed loop system; sustainable energy systems; task scheduling.
Community detection of trajectory data for location-based facility recommendation system
by B.A. Sabarish, R. Karthi, T. Gireesh Kumar
Abstract: Trajectory contains spatial-data generated from traces of moving objects like people, animals, etc. Community generated from trajectories portrays common behaviour. Trajectory clustering based on community-detection involves region-graph generation and community-detection. In region-graph generation, Trajectories are projected to spatial grid to transform GPS representation into string representation. Sequential graph is generated from string representation. Edge-based similarity is calculated between trajectories to create an adjacency matrix representing relationship and represent entire region. In community-detection phase, region-graph is divided into communities using various algorithms and validated using modularity values. Based on analysis, Louvain, fast-greedy, leading-eigenvector, and edge-Betweenness algorithms provide the optimum modularity value for better community detection. Analysing the community can be used as a pre-processing step in identifying location for location-based services (LBS), including hotspots, delay-tolerant-networks, and mobile antenna placements for better coverage. Design and capacity planning of the network based on the size and pattern of the community improves quality of LBS.
Keywords: trajectory; community; delay tolerant networks; quality of service; clustering; representation.
Information integration method of English teaching resources based on artificial intelligence
by Jin Guo
Abstract: In this paper, an information integration method of English teaching resources based on artificial intelligence is proposed. The generalised fuzzy C-means clustering algorithm was used to construct the Arduino device image dataset, and the convolutional neural network model of Arduino device was designed. The image data input model in TFRecord format was designed, and multiple Arduino device resource feature maps were output through convolution, pooling and other operations to establish the English teaching resource library and complete the information integration of English teaching resources. The experimental results show that this method has fast convergence speed, with a recognition success rate of 96.7% and can improve the academic performance to more than 90 points, and the actual evaluation value of it in the 5th and 6th academic year is close to 1. Therefore, it can improve the information integration efficiency of English teaching resources and English performance.
Keywords: artificial intelligence; English teaching; resource information integration; Arduino device; convolution neural network; clustering algorithm.
Research on reversible information hiding in image encryption domain based on multilayer perceptron
by Zhiqiang Yue, Weijia Chai
Abstract: To solve the problems of large mean square error, low peak signal-to-noise (PSNR) ratio and low embedding rate of traditional methods, a reversible information hiding in image encryption domain based on multilayer perceptron is proposed. Image blocks are encrypted sequentially, and the encrypted information is embedded into the image to complete the construction of the image encryption domain. Reversible information in the image encryption domain is extracted, the weight of the extracted information is calculated by multilayer perceptron, and the sender and receiver models are built according to the calculation results. Reversible information hiding in the image encryption domain is realised by using these two models. Experimental results show that the maximum and minimum mean square error of the encrypted image and the original image are 0.254 and 0.482 respectively, the maximum and minimum PSNR ratio are 56dB and 50dB respectively, and embedding rate is always above 91%.
Keywords: multilayer perceptron; images; encrypted domain; reversible information hiding; dimensionality reduction; sender model; receiver model.
A key feature mining method of online teaching behavior based on k-kernel decomposition
by Wei Wang
Abstract: In view of the poor effect of online teaching behaviour key feature mining, an online teaching behaviour key feature mining method based on k-kernel decomposition is designed. Firstly, the adjacent data of the key features of network teaching behaviour are interpolated to determine the key features, and the singular distance function is normalised to complete the feature preprocessing; Then, the key characteristics of network teaching behaviour are transformed into weighted network, and the key characteristics are divided according to the centrality of nodes; Finally, the online behaviour feature k-kernel after classification is assigned, the feature k-kernel value index after assignment is calculated, the correlation of feature data is calculated, the probability of feature data belonging to k-clustering is determined, and the key feature mining of network teaching behaviour is completed. The results show that the mining effect of this method is good.
Keywords: K-kernel decomposition; weighted network; online teaching behaviour; key feature mining.
A colour transfer method of interior design based on machine learning
by Tiesheng Liu
Abstract: In order to overcome the problems of large colour range, low structure similarity and poor objective evaluation index in the process of interior design colour transfer, this paper proposes a new method of interior design colour transfer based on machine learning. In this method, K-means algorithm is introduced to eliminate the uneven brightness area of the target image. The target image is processed by initial clustering, and the iterative threshold segmentation method is used to obtain the final clustering accurate target image. Combined with machine learning, the corresponding samples are selected on the two images to finish colouring, and the colour transfer of interior design is realised by referring to the coloured sample block. The experimental results show that the colour degree of the proposed method is maintained between 33 and 45, the structural similarity is always above 95%, and the comprehensive objective evaluation index value is close to 100%.
Keywords: machine learning; interior design; colour transfer; K-means algorithm.
Colour image reconstruction in indoor space based on machine vision
by Zhang Na
Abstract: In order to solve the problems of low accuracy and long time consuming in colour image reconstruction in indoor space, a colour image reconstruction method in indoor space based on machine vision is proposed. Through obtaining the maximum and minimum grey value of colour image in indoor space, the colour image features are extracted; according to the brightness contrast of colour image channel, the brightness gain value is set; the colour base colour and the initial image are mixed to enhance the colour image in indoor space; the SDF value of the colour image in indoor space is obtained by using the voxel mapping method of machine vision algorithm, and the overlapping parts of each frame of the image are fused to realise the colour image reconstruction in indoor space. The experimental results show that the highest accuracy of the proposed method is 98%.
Keywords: machine vision; indoor space colour; image reconstruction; image enhancement.
Research on financial information disclosure risk management method based on internet of things
by Ximeng Li
Abstract: In order to overcome the problem that the traditional risk management method has difficulty achieving the expected effect due to the careless classification of risk levels, this paper designs the financial information disclosure risk management method based on the internet of things. Based on the network financial information disclosure traceability model, the risk assessment index system is built, and the risk events are classified by ranking the risk events. The evaluation result is divided into five categories and the risk warning level is set. Then, the wireless sensor technology in the internet of things is used to design data fusion algorithm and distributed routing algorithm to achieve risk management. Simulation results show that the recall rate of the risk assessment index is above 94%, and the time to issue risk warning is always less than 6 min, which fully proves the effectiveness of the method.
Keywords: internet of things; financial information disclosure; wireless sensor technology; disclosure traceability mode; risk perception; risk assessment; risk warning level; fine division of grades.
Spatial texture feature classification algorithm for high resolution 3D images
by Ping Wang
Abstract: The existing feature classification algorithms have a lot of noise in the process of classification, which leads to the problems of low classification efficiency and unbalanced classification of image spatial texture features. Based on this, a texture feature classification algorithm based on RUSBoost is proposed. Wavelet coefficients, threshold processing and image reconstruction are used to denoise the image. On the basis of BDAWPSO algorithm, image segmentation is carried out by searching for the optimal threshold. Gabor transform and windowing are used to overcome the lack of local analysis ability and reduce the classification time. The original unbalanced image data is converted into new balanced data by using Rus boost algorithm. The experimental results show that the algorithm can improve the classification effect and display the texture information of the image better.
Keywords: high resolution; 3D image; spatial texture feature; classification algorithm.
Research on interactive data packet storage algorithm for Hadoop cloud computing platform
by Yu-feng Ou, Yan-Xi Li, Wei-Guo Liu
Abstract: The accuracy of traditional interactive data packet storage algorithms for data packet storage is not high, the storage time is very large, and the integrity of the data cannot be guaranteed. Therefore, a new type of interactive data packet storage algorithm is proposed. First, Bayes' theorem is used to calculate the frequency of grouping categories, so as to construct the interactive data classification model of the Hadoop cloud computing platform. By using the relationship between Smart-DIRPE codes and rules, the integrity of data packet storage is effectively ensured, and the domain conversion method is used to realise the range matching design of interactive data packets. Based on this, the design of the interactive data packet storage algorithm is completed. Experimental results show that the algorithm has higher grouping accuracy, better data integrity and shorter time consumption. When the data size is 90 MB, the time consumption is 2.3 s.
Keywords: Hadoop cloud computing platform; interactive data; packet storage.
Construction of accounting information distortion evaluation model based on artificial intelligence technology
by Qingmin Yu
Abstract: In order to overcome the problems of low accuracy of the current accounting information distortion evaluation methods, this paper designs an accounting information distortion evaluation model based on artificial intelligence technology. This paper analyses the causes, types, specific characteristics and performance of accounting information distortion, focuses on the statistics of the characteristics of accounting information distortion, constructs the initial evaluation index system according to the statistical analysis results, and selects the principal component indicators by principal component analysis. Through the establishment of classification and regression tree pruning, combined with artificial intelligence technology to build accounting information distortion evaluation model, completes the accurate evaluation of accounting information distortion. The simulation results show that the evaluation error of the model is between 0%-1.1%, and the economic loss to the enterprise is about 15,100 yuan, which can reduce the evaluation error and the economic loss of the enterprise.
Keywords: artificial intelligence technology; accounting information distortion; evaluation model; principal component analysis; PCA; classification regression tree.
Research on fault signal detection method of mechanical vibration based on Kalman filtering algorithm
by Yaozeng Jia, Ye Hu
Abstract: There are some problems in traditional mechanical fault signal detection, such as large fault detection error and is time-consuming. A mechanical vibration fault signal detection method based on Kalman filter algorithm is proposed. The time-varying nonlinear oil film force of mechanical equipment under different working conditions is analysed. The piezoelectric acceleration sensor is set in the mechanical equipment to obtain the mechanical vibration fault signal. The energy difference between the normal signal and the fault signal is obtained by arranging them according to the acquisition time sequence. The mechanical vibration fault signal is pre-processed by Kalman filtering algorithm to obtain the state prediction of the fault signal. The frequency and amplitude characteristics of mechanical vibration fault signal are obtained to complete the detection of mechanical vibration fault signal. The results show that the error of the proposed method is small and the detection time is 2 s.
Keywords: Kalman filtering algorithm; mechanical vibration; vibration fault signal; signal detection.
Research on target classification method for dense matching point cloud based on improved random forest algorithm
by Tiebo Sun, Jinhao Liu, Jiangming Kan, Tingting Sui
Abstract: Aiming at the problems of low accuracy and low efficiency of traditional point cloud target classification methods, this paper designs a new classification method based on improved random forest algorithm. Bagging is combined with random subspace to form a subset of feature training at random, so that the generalisation ability of random forest algorithm can be increased while the data processing speed can be accelerated to avoid overfitting phenomenon. On the basis of extracting geometric features of coloured point clouds, the optimal feature subset for classification is determined, and then the dense matching point clouds are classified using the improved random forest algorithm. Experimental results show that the classification error rate of this method is less than 1%, the average classification process takes only 83.995 s, and the VIM value is all over 0.1, indicating that this method can effectively improve the classification effect of dense matching point cloud targets.
Keywords: improved random forest algorithm; dense matching point cloud; target classification; optimal feature subset.
Design and implementation of smart power firefighting IoT based on massive heterogeneous terminals
by Zhili Ma, Feng Wei, Xi Song, Zhicheng Ma
Abstract: This paper is mainly aimed at a large number of old, unintelligent power system terminals, a set of intelligent IoT framework solutions with overall coordination, linkage, and perception have been researched and designed to solve the real-time monitoring of various terminal abnormalities and faults in the unattended situation. The specific methods here are as follows. First, a status monitor to the back end of the old equipment is added for grasping the operational health index of each terminal in real time. Secondly, the thermal imaging module, the fire water system module, and the fire alarm module are introduced to build a power fire monitoring network that can cover the entire area. Then, on the basis of setting thresholds for each monitoring module, advanced information network processing technology is used to realise fault warning for each monitoring object in the area. Finally, a statistical analysis module is embedded in the background to perform real-time statistics on various fire protection malfunctions and monitor the completion rate of each unit's failure rectification.
Keywords: the power systems; smart internet of things; real-time perception; fault warning.
Study on the detection and recovery algorithm of important financial information tampering
by Wenjin Wang, Xinghua Liu, Yuxin Zhang, Jia Liu, Ping Jiang
Abstract: A self-recovery watermarking algorithm for important financial information content authentication is proposed in this paper. The algorithm firstly pre-processes the source image with wavelet transform and divides sub-band image for low frequency into sub-blocks. Secondly, in order to ensure the accuracy of the localisation of the tampered area, a dual authentication watermarking mechanism called 'SVD-position recording' is adopted for each image block. Thirdly, the source images are down-scaled and segmented based on K-mediods clustering to generate a binary code stream of vital information, taking into account the watermark capacity and recovery quality. Finally, the binary code stream is knight's parade scrambled and RS error-corrected coded to be embedded in the high-frequency wavelet coefficients which reduce the possibility of recovery watermark tampering. Experimental results show that the peak signal-to-noise ratio (PSNR) of the financial document images with embedded watermark by the algorithm in this paper is above 43 dB, which has excellent visual quality. At the same time, it achieves accurate localisation and precise recovery for tampering attacks such as add, copy, and delete, which is suitable for tampering detection and recovery of important financial information.
Keywords: financial information security; financial document image; tamper detection; self-recovery watermarks.
Special Issue on: Human Computer Interaction for Speech and Augmentative Communication
Brain computer interface with EEG signals to improve feedback system in higher education based on augmentative and alternative communication
by Hua Sun, Hongxia Hou, Wenxia Song, Hongjuan Hu, Na Liu, Achyut Shankar
Abstract: A predictive brain-computer interface is a way to monitor EEG signals in humans currently under the experiment to understand the capability and get proper feedback about the digital education systems. The use of dynamically processed and collected data in a feedback system is unfeasible. Substantial processing delays are caused by a large volume of data utilised by the modern higher educational ideas. An artificial intelligence assisted brain-computer interface feedback system (AI-BCIFS) for ugmentative and alternative communication is proposed to improve feedback analysis based on the EEG signals. AI-BCIFS method is proposed to avoid an unwanted and improper understanding of feedback in the higher education systems. An expanded optimisation methodology is introduced based on the feedback analysis, and the enhanced seeking feedback protocol (ESFP) has been developed to describe automatic recognition and storage. The experimental studies show that AI-BCIFAC is preferable to the existing approaches in terms of accuracy.
Keywords: brain-computer interface; BCI; higher education; signal; augmentation.
A deep learning approach in brain-computer interaction for augmentative and alternative communication
by Yubin Liu
Abstract: The electroencephalography classification is the primary aspect of brain-computer systems. The changes there are two main concerns. Firstly, conventional approaches do not use multimodal knowledge to their fullest degree. Second, it is almost unlikely to obtain the rule-based EEG repositories as genetic information processing is complex, and metadata accuracy is expensive. In this sense, researchers suggest a new approach named deep learning-based brain-computer interaction (DLBCI) for augmentative and alternative communication to profound transfer learning to address these issues. Initially, these model perceptual activities based on EEG signals, using the EEG input images characterisation, which is intended to retain a standardised description of multimodal ECG signals. Secondly, this model develops a deep-scope transmission of knowledge through joint operations, including an opponents infrastructure and an incredibly unique transfer function. The proposed framework for EEG classification problems, such as strength and precision, has many economic benefits in experimentation.
Keywords: deep learning; brain-computer interaction; augmentative and alternative communication; multimodal signal.
EFL teaching based on big data analytics in mobile AAC system
by Junsheng Li
Abstract: Augmentative and alternative communication, the mechanisms that may complement or substitute formal language in certain situations, tends to lead to technological assistance in English as a foreign language (EFL). In this context, this paper discusses the mobile AAC concept, which can enhance EFL teaching performance. The article proposes a hybrid approach to EFL using big data analytics and mobile AAC (EFLAAC) to facilitate smooth access for mobile learners to educational digital content on various mobile devices. The compression and decompression program in the new framework compress the file during the streaming without changing the quality. A systematic subjective analysis has been carried out with dataset 0f 250. The proposed model is evaluated and delivers high learning results from multimedia instructional videos using pre and post-test questionnaires. EFLAAC attains 94.9% of accuracy, which is comparatively good than other approaches.
Keywords: EFL teaching; mobile augmentative; alternative communication; big data analytics; digital learning.
Improving college ideological and political education based on deep learning
by Youwu Zhang, Yongquan Yan, R. Lakshmana Kumar, Sapna Juneja
Abstract: The rapid development of information and technology results in the involvement of technology channels like communication devices, and simultaneously it acts as a vital part of life. It emerged as a significant concern in the students educational progress both physically and mentally. It is essential to maintain the teaching quality in the teaching field, and more concentration is needed for the college ideological political education. For successive enhancement, a novel multimedia assisted ideological and political education system using deep learning techniques (MIPE-DLT) is introduced. The model analyses the characteristics and the capability of higher education students in gathering information and realising the effects of propagating novelties in ideological and political education. The proper flow of protocols has been executed in implementing multimedia techniques towards ideological and political education. It bridges the gap efficiently with a higher accuracy rate and processing rate. Compared with previous techniques, the MIPE-DLT achieves a high-order performance ratio with a minimal delay rate.
Keywords: multimedia; political; ideology; education; college learning; quality; teaching.