ISSN 
2021 Vol. 50, No. 3
2021, 50(3): 321-331.
doi: 10.12178/1001-0548.2021046
Abstract:
For the different kinds of packet transmission requirements in vehicular ad-hoc network (VANET), this paper proposes a hybrid message transmission mechanism and routing algorithm combining vehicle to vehicle (V2V) and 5G cellular communications. The packets in VANETs are divided into two types: delay sensitive and delay insensitive. The delay-sensitive packets can be transmitted efficiently taking the advantages of 5G cellular network with low delay, high reliability and wide network coverage. Since the ad-hoc network has a lower cost than the 5G cellular network, a routing algorithm based on bus assistance is designed for delay insensitive data packets. In order to reduce the probability of communication link disconnection, a prediction mechanism for the update-time of the neighbor table is proposed. The effectiveness of the scheme is verified by a large number of simulation experiments. Simulation results show that the proposed routing scheme can improve the delivery ratio of packets, reduce the end-to-end delay and decrease end-to-end transmission hops.
For the different kinds of packet transmission requirements in vehicular ad-hoc network (VANET), this paper proposes a hybrid message transmission mechanism and routing algorithm combining vehicle to vehicle (V2V) and 5G cellular communications. The packets in VANETs are divided into two types: delay sensitive and delay insensitive. The delay-sensitive packets can be transmitted efficiently taking the advantages of 5G cellular network with low delay, high reliability and wide network coverage. Since the ad-hoc network has a lower cost than the 5G cellular network, a routing algorithm based on bus assistance is designed for delay insensitive data packets. In order to reduce the probability of communication link disconnection, a prediction mechanism for the update-time of the neighbor table is proposed. The effectiveness of the scheme is verified by a large number of simulation experiments. Simulation results show that the proposed routing scheme can improve the delivery ratio of packets, reduce the end-to-end delay and decrease end-to-end transmission hops.
2021, 50(3): 339-346.
doi: 10.12178/1001-0548.2020384
Abstract:
In this paper, based on the cubic polynomial relation between charge q and magnetic flux\begin{document}$ \varphi$\end{document} ![]()
![]()
In this paper, based on the cubic polynomial relation between charge q and magnetic flux
2021, 50(3): 347-353.
doi: 10.12178/1001-0548.2021012
Abstract:
With the demand of satellite communication system for variable and efficient payloads, space traveling-wave tube (TWT) amplifier will be accelerated from single dedicated mode to power flexible multi-tasking mode. In order to solve the design difficulties of high efficiency power flexible space TWT under different output power conditions, a Ku-band helix radiation cooled space TWT has been studied. By improving the electron efficiency, collector efficiency and electron transmission rate under different anode voltages, in the 500 MHz bandwidth, the Ku-band TWT can generate continuous-wave output power over 150W and the overall efficiency greater than 68%. And the overall efficiency of the TWT is more than 60% when the output power back-off 3dB.
With the demand of satellite communication system for variable and efficient payloads, space traveling-wave tube (TWT) amplifier will be accelerated from single dedicated mode to power flexible multi-tasking mode. In order to solve the design difficulties of high efficiency power flexible space TWT under different output power conditions, a Ku-band helix radiation cooled space TWT has been studied. By improving the electron efficiency, collector efficiency and electron transmission rate under different anode voltages, in the 500 MHz bandwidth, the Ku-band TWT can generate continuous-wave output power over 150W and the overall efficiency greater than 68%. And the overall efficiency of the TWT is more than 60% when the output power back-off 3dB.
2021, 50(3): 354-359.
doi: 10.12178/1001-0548.2020165
Abstract:
Aiming at the problem that the generalized spatial modulation (GSM) system can not take into account the high channel gain and low correlation degree under correlated channels and the applicable scenarios of antenna selection technology, an incremental antenna combination selection algorithm based on norm and correlation is proposed. First the norms of all antenna combinations are calculated and sorted. Secondly, the antenna combination with the largest norm is selected and added into the candidate set. Finally, the antenna combination with the lowest degree of correlation with the antenna combination in the candidate set is selected from the remaining set and added to the candidate set one by one. The algorithm uses the effective channel vector as a quantitative indicator to evaluate whether the antenna combination is excellent. The angle among the effective channel vectors not only explicitly reflects the correlation among the antenna combinations, but also implicitly reflects the range of the channel vector. While maximizing the channel gain, the degree of identification among the combinations is improved. Simulation shows that when the bit error rate is 10−3, the algorithm can obtain a gain of 2 dB compared to the random selection of antenna combination algorithm. Compared with the receiving antenna selection algorithm under the same spectrum efficiency, this algorithm obtains a gain of about 1.3 dB.
Aiming at the problem that the generalized spatial modulation (GSM) system can not take into account the high channel gain and low correlation degree under correlated channels and the applicable scenarios of antenna selection technology, an incremental antenna combination selection algorithm based on norm and correlation is proposed. First the norms of all antenna combinations are calculated and sorted. Secondly, the antenna combination with the largest norm is selected and added into the candidate set. Finally, the antenna combination with the lowest degree of correlation with the antenna combination in the candidate set is selected from the remaining set and added to the candidate set one by one. The algorithm uses the effective channel vector as a quantitative indicator to evaluate whether the antenna combination is excellent. The angle among the effective channel vectors not only explicitly reflects the correlation among the antenna combinations, but also implicitly reflects the range of the channel vector. While maximizing the channel gain, the degree of identification among the combinations is improved. Simulation shows that when the bit error rate is 10−3, the algorithm can obtain a gain of 2 dB compared to the random selection of antenna combination algorithm. Compared with the receiving antenna selection algorithm under the same spectrum efficiency, this algorithm obtains a gain of about 1.3 dB.
2021, 50(3): 360-367.
doi: 10.12178/1001-0548.2020053
Abstract:
Based on sparse and redundant representation, parallel coordinate descent (PCD) is one of the best denoising algorithms. In audio signal processing, however, when the number of segmented frames is large, the computational burden is heavy. Processing each frame separately with the PCD algorithm causes a dramatic increase in time cost. Therefore, this paper proposes a new time-domain framework and a heuristic joint PCD (called joint-PCD) algorithm based on joint sparse representation (JSR) and simultaneous sparse approximation (SSA). In this framework, each audio frame is used as a column vector to generate a signal matrix. Utilizing an over-complete dictionary, joint-PCD is used to synchronously (simultaneously) denoise a signal matrix (instead of an audio frame), which greatly improves the efficiency of the algorithm and reduces the burden of running time. The simulation results show that the joint-PCD algorithm not only has almost the same and excellent denoising performance with PCD, but also increases the denoising speed of PCD by about five times, which greatly improves the convergence performance of the PCD algorithm.
Based on sparse and redundant representation, parallel coordinate descent (PCD) is one of the best denoising algorithms. In audio signal processing, however, when the number of segmented frames is large, the computational burden is heavy. Processing each frame separately with the PCD algorithm causes a dramatic increase in time cost. Therefore, this paper proposes a new time-domain framework and a heuristic joint PCD (called joint-PCD) algorithm based on joint sparse representation (JSR) and simultaneous sparse approximation (SSA). In this framework, each audio frame is used as a column vector to generate a signal matrix. Utilizing an over-complete dictionary, joint-PCD is used to synchronously (simultaneously) denoise a signal matrix (instead of an audio frame), which greatly improves the efficiency of the algorithm and reduces the burden of running time. The simulation results show that the joint-PCD algorithm not only has almost the same and excellent denoising performance with PCD, but also increases the denoising speed of PCD by about five times, which greatly improves the convergence performance of the PCD algorithm.
2021, 50(3): 368-374.
doi: 10.12178/1001-0548.2020161
Abstract:
To deal with the slow operation speed in the existing single-precision floating-point multiplier, a fast Karatsuba-based single-precision floating-point multiplier which combines the advantages of Karatsuba algorithm with the Vedic algorithm is designed in this paper. The fast Karatsuba-based multiplier decreases the multiplication-operation times of the conventional single-precision floating-point multiplier by splitting the multiplication of 24-bit mantissa into that of fewer mantissa. An improved multiplication architecture composed of the 3-bit and 4-bit mantissa is constructed and further optimized by employing the Vedic algorithm. The 3-bit and 4-bit multipliers are respectively achieved by the corresponding adders with low complexity and fast speed, leading to faster processing speed. The results of simulation and FPGA verification imply that the designed single-precision floating-point multiplier achieves approximately 5 times and 2 times higher performance in the maximum operating clock frequency, comparing to the conventional Karatsuba-based and the Vedic-based single-precision floating-point multiplier, respectively.
To deal with the slow operation speed in the existing single-precision floating-point multiplier, a fast Karatsuba-based single-precision floating-point multiplier which combines the advantages of Karatsuba algorithm with the Vedic algorithm is designed in this paper. The fast Karatsuba-based multiplier decreases the multiplication-operation times of the conventional single-precision floating-point multiplier by splitting the multiplication of 24-bit mantissa into that of fewer mantissa. An improved multiplication architecture composed of the 3-bit and 4-bit mantissa is constructed and further optimized by employing the Vedic algorithm. The 3-bit and 4-bit multipliers are respectively achieved by the corresponding adders with low complexity and fast speed, leading to faster processing speed. The results of simulation and FPGA verification imply that the designed single-precision floating-point multiplier achieves approximately 5 times and 2 times higher performance in the maximum operating clock frequency, comparing to the conventional Karatsuba-based and the Vedic-based single-precision floating-point multiplier, respectively.
2021, 50(3): 375-381.
doi: 10.12178/1001-0548.2019303
Abstract:
The coexistence mechanism of LTE-U (long term evolution - unlicensed) and WiFi in 5G heterogeneous wireless networks is discussed. Q learning based – almost blank subframe (QL-ABS) configuration mechanism is proposed for LTE-U. In the mechanism, the queuing theory is used to model the heterogeneous network where LTE-U and WiFi coexist, and the network delay performance is used to represent the input state of Q-learning. The simulation results show that the proposed mechanism can generate a more reasonable almost blank subframe configuration strategy for LTE-U under multiple services and different load conditions through autonomous learning process, and therefore has better online learning performance. Compared with the traditional methods, the proposed mechanism better solves the problem of coexistence between LTE-U and existing WiFi network in the unlicensed frequency bands, and improves the overall delay and online performances of the networks.
The coexistence mechanism of LTE-U (long term evolution - unlicensed) and WiFi in 5G heterogeneous wireless networks is discussed. Q learning based – almost blank subframe (QL-ABS) configuration mechanism is proposed for LTE-U. In the mechanism, the queuing theory is used to model the heterogeneous network where LTE-U and WiFi coexist, and the network delay performance is used to represent the input state of Q-learning. The simulation results show that the proposed mechanism can generate a more reasonable almost blank subframe configuration strategy for LTE-U under multiple services and different load conditions through autonomous learning process, and therefore has better online learning performance. Compared with the traditional methods, the proposed mechanism better solves the problem of coexistence between LTE-U and existing WiFi network in the unlicensed frequency bands, and improves the overall delay and online performances of the networks.
2021, 50(3): 382-390.
doi: 10.12178/1001-0548.2021004
Abstract:
Localization is one of the core technologies for mobile robots to achieve full autonomous movement and is a prerequisite for other autonomous tasks. The robot working environment is dynamic in most cases, so the localization algorithm must overcome the effects of dynamic changes in the environment. The paper proposes a localization algorithm that allows the robot to perform robust and life-long localization in dynamic environment. The algorithm can not only filter out high-dynamic objects but also update semi-static object on the map at the same time, and it can also use the information provided in semi-static objects to improve localization performance. In this paper, the processing of dynamic objects is divided into two parts: filtering of high-dynamic objects and updating of semi-static objects. For high dynamic object filtering, a dynamic object detection method combining a delay comparison method and a tracking method is proposed by observing the characteristics of localization system; for the update of semi-static objects, this paper uses the pose graph optimization and occupancy map to implement the dynamic update of the map. The combination of the two methods allows the robot to achieve long-term stable localization in a dynamic environment. The experimental results demonstrate that the proposed method allows the robot to achieve long-term localization, overcome the effects of high-dynamic objects and keep the map always consistent with the environment.
Localization is one of the core technologies for mobile robots to achieve full autonomous movement and is a prerequisite for other autonomous tasks. The robot working environment is dynamic in most cases, so the localization algorithm must overcome the effects of dynamic changes in the environment. The paper proposes a localization algorithm that allows the robot to perform robust and life-long localization in dynamic environment. The algorithm can not only filter out high-dynamic objects but also update semi-static object on the map at the same time, and it can also use the information provided in semi-static objects to improve localization performance. In this paper, the processing of dynamic objects is divided into two parts: filtering of high-dynamic objects and updating of semi-static objects. For high dynamic object filtering, a dynamic object detection method combining a delay comparison method and a tracking method is proposed by observing the characteristics of localization system; for the update of semi-static objects, this paper uses the pose graph optimization and occupancy map to implement the dynamic update of the map. The combination of the two methods allows the robot to achieve long-term stable localization in a dynamic environment. The experimental results demonstrate that the proposed method allows the robot to achieve long-term localization, overcome the effects of high-dynamic objects and keep the map always consistent with the environment.
2021, 50(3): 391-397.
doi: 10.12178/1001-0548.2021025
Abstract:
In this paper, a multi-sensor fusion technology for micro air vehicle formation crossing obstacles under low-precision GPS conditions is proposed based on the monocular vision positioning method, altimetry filter algorithm based on LiDAR, and the leading-following formation strategy of micro air vehicle. For verifying this technology, a complete set of the micro air vehicle formation test platform was established through the design of key modules such as the hardware architecture of the system, sensor distribution, airborne self-organizing network, and the software interface of the airborne formation ground station. The actual formation flying test of the outdoor test site was conducted on the base of the simulation verification of the visual positioning algorithm, time-domain filtering of laser data, and formation flying under the robot operating system. Finally, formation flying missions such as formation flying across multiple groups of door frame obstacles at random positions, online path planning, and one-key take-off and return were realized using this technology, verifying the feasibility of the technology in the outdoor complex flight environment.
In this paper, a multi-sensor fusion technology for micro air vehicle formation crossing obstacles under low-precision GPS conditions is proposed based on the monocular vision positioning method, altimetry filter algorithm based on LiDAR, and the leading-following formation strategy of micro air vehicle. For verifying this technology, a complete set of the micro air vehicle formation test platform was established through the design of key modules such as the hardware architecture of the system, sensor distribution, airborne self-organizing network, and the software interface of the airborne formation ground station. The actual formation flying test of the outdoor test site was conducted on the base of the simulation verification of the visual positioning algorithm, time-domain filtering of laser data, and formation flying under the robot operating system. Finally, formation flying missions such as formation flying across multiple groups of door frame obstacles at random positions, online path planning, and one-key take-off and return were realized using this technology, verifying the feasibility of the technology in the outdoor complex flight environment.
2021, 50(3): 398-404.
doi: 10.12178/1001-0548.2021015
Abstract:
In the offloading problem of mobile edge computing task, the traditional offloading algorithm only considers the computing resources of mobile devices and edge servers, and has some limitations in resource utilization and system efficiency. this paper proposes an edge-cloud weighted serial task offloading algorithm based on rainbowDQN (ECWS-RDQN) based on the RainbowDQN algorithm, considering the factors of delay, energy consumption cost and service quality assurance. This algorithm realizes the serial task dynamic assignment processing of network edge and cloud collaboration through the weight to provide approximately optimal task assignment offloading strategies for different user device applications. Experiments show that the ECWS-RDQN algorithm has better system efficiency than the traditional schemes and improves the service quality of the applications.
In the offloading problem of mobile edge computing task, the traditional offloading algorithm only considers the computing resources of mobile devices and edge servers, and has some limitations in resource utilization and system efficiency. this paper proposes an edge-cloud weighted serial task offloading algorithm based on rainbowDQN (ECWS-RDQN) based on the RainbowDQN algorithm, considering the factors of delay, energy consumption cost and service quality assurance. This algorithm realizes the serial task dynamic assignment processing of network edge and cloud collaboration through the weight to provide approximately optimal task assignment offloading strategies for different user device applications. Experiments show that the ECWS-RDQN algorithm has better system efficiency than the traditional schemes and improves the service quality of the applications.
2021, 50(3): 405-413.
doi: 10.12178/1001-0548.2020359
Abstract:
Collaborative filtering techniques require tremendous amount of personal data to provide personalized recommendation services, which has caused the rising concerns about the risk of privacy leakage. Most existed methods for implementing privacy protection in recommender systems are prone to introduce excessive noises, which significantly degrades the recommendation quality. To address this problem, a matrix factorization algorithm satisfying differential privacy is proposed. The method first converts the matrix factorization problem into two alternate optimization problems, in which user latent factors and item latent factors are optimized respectively. Then a genetic algorithm is introduced to solve these two optimization problems, in which the enhanced exponential mechanism is applied into the individual selection and a novel mutation operation is designed based on the idea of finding important latent factors. Theoretical analysis and experimental results show that the algorithm can not only provide strong differential privacy protection for user data, but also ensure the accuracy of recommendations. Therefore, it has good application value in recommender systems.
Collaborative filtering techniques require tremendous amount of personal data to provide personalized recommendation services, which has caused the rising concerns about the risk of privacy leakage. Most existed methods for implementing privacy protection in recommender systems are prone to introduce excessive noises, which significantly degrades the recommendation quality. To address this problem, a matrix factorization algorithm satisfying differential privacy is proposed. The method first converts the matrix factorization problem into two alternate optimization problems, in which user latent factors and item latent factors are optimized respectively. Then a genetic algorithm is introduced to solve these two optimization problems, in which the enhanced exponential mechanism is applied into the individual selection and a novel mutation operation is designed based on the idea of finding important latent factors. Theoretical analysis and experimental results show that the algorithm can not only provide strong differential privacy protection for user data, but also ensure the accuracy of recommendations. Therefore, it has good application value in recommender systems.
2021, 50(3): 414-421.
doi: 10.12178/1001-0548.2020403
Abstract:
To improve the accuracy of COVID-19 trend forecasting, a method of COVID-19 trend forecasting by using dropout and long short-term memory (LSTM) is proposed. The method uses web crawler based on python to obtain complete domestic historical data of COVID-19, which improves the efficiency of data collection and reduces data errors caused by subjective reasons. To avoid adding time features artificially and explore the nonlinear relationship fully between the less data of COVID-19, the proposed model extends the layers of the deep learning network. Then, the dropout technique is applied to the non-circular part of each hidden layer to randomly deactivate neurons, preventing the neural network from overfitting. The experiment demonstrates that the method can predict the number of cumulative confirmed cases, current confirmed cases and recovered cases.
To improve the accuracy of COVID-19 trend forecasting, a method of COVID-19 trend forecasting by using dropout and long short-term memory (LSTM) is proposed. The method uses web crawler based on python to obtain complete domestic historical data of COVID-19, which improves the efficiency of data collection and reduces data errors caused by subjective reasons. To avoid adding time features artificially and explore the nonlinear relationship fully between the less data of COVID-19, the proposed model extends the layers of the deep learning network. Then, the dropout technique is applied to the non-circular part of each hidden layer to randomly deactivate neurons, preventing the neural network from overfitting. The experiment demonstrates that the method can predict the number of cumulative confirmed cases, current confirmed cases and recovered cases.
2021, 50(3): 422-427.
doi: 10.12178/1001-0548.2021062
Abstract:
Link prediction is a significant and challenging task in network science, which plays an important role in friend recommendations, the discovery of biological interactions, link navigation, and product recommendations. Since the rise of link prediction, many methods have been proposed. Due to the simplicity, interpretability, high efficiency, scalability, and satisfactory performance, local similarity indices are widely used in various research fields and applications. Under the 2-hop-based neighborhood analytical framework, most of the indices are proposed based on the network organization mechanisms including homophily, clustering and triadic closure. In the last decade, the emergence of local community paradigm, Hebb theory and new arguments about the rationality of the 2-hop-based framework has greatly promoted the development of local similarity indices. This paper aims at sorting out and discussing these new findings.
Link prediction is a significant and challenging task in network science, which plays an important role in friend recommendations, the discovery of biological interactions, link navigation, and product recommendations. Since the rise of link prediction, many methods have been proposed. Due to the simplicity, interpretability, high efficiency, scalability, and satisfactory performance, local similarity indices are widely used in various research fields and applications. Under the 2-hop-based neighborhood analytical framework, most of the indices are proposed based on the network organization mechanisms including homophily, clustering and triadic closure. In the last decade, the emergence of local community paradigm, Hebb theory and new arguments about the rationality of the 2-hop-based framework has greatly promoted the development of local similarity indices. This paper aims at sorting out and discussing these new findings.
2021, 50(3): 428-436.
doi: 10.12178/1001-0548.2020350
Abstract:
The sequence of authorship in academic papers provides important information for the study of their academic contributions. It is a significant research topic in the field of science of science to determine the important author of the article according to the author’s signature positions, such as the chaperone effect proposed in recent years. The paper analyzes the differences among authors in different positions, defines the first author, second author, third author, last author, corresponding author and penultimate author in different signature positions as the important authors of the article to detect the chaperone effect systematically. The results show the generalized chaperone effect. At the same time, the paper analyzes the change trend and the differences of the chaperone effect of authors in different positions. We explain the differences by calculating the similarity coefficient among authors in different positions and corresponding authors. When there is a high coincidence among authors in different signature positions and corresponding authors, such as the coincidence between the last authors and corresponding authors in comprehensive journals is higher, the coincidence between the first authors and corresponding authors in comprehensive journals is higher. This study can help us to understand the role and the contribution of authors in academic articles, and further develop new scientific theories based on author’s signature positions.
The sequence of authorship in academic papers provides important information for the study of their academic contributions. It is a significant research topic in the field of science of science to determine the important author of the article according to the author’s signature positions, such as the chaperone effect proposed in recent years. The paper analyzes the differences among authors in different positions, defines the first author, second author, third author, last author, corresponding author and penultimate author in different signature positions as the important authors of the article to detect the chaperone effect systematically. The results show the generalized chaperone effect. At the same time, the paper analyzes the change trend and the differences of the chaperone effect of authors in different positions. We explain the differences by calculating the similarity coefficient among authors in different positions and corresponding authors. When there is a high coincidence among authors in different signature positions and corresponding authors, such as the coincidence between the last authors and corresponding authors in comprehensive journals is higher, the coincidence between the first authors and corresponding authors in comprehensive journals is higher. This study can help us to understand the role and the contribution of authors in academic articles, and further develop new scientific theories based on author’s signature positions.
2021, 50(3): 437-445.
doi: 10.12178/1001-0548.2020401
Abstract:
In the era of information explosion, online social network has been widely depended and applied as the main way of information transmission. However, the dynamic process of information transmission is often difficult to accurately predict and prevent in online social network. In this paper, the hyperedges in hypergraphs are introduced to describe complex social relationships between two or more individuals. Based on the hypernetwork dynamic evolution model, we construct online social hypernetwork, and combine with the susceptible infected susceptible (SIS) model based on the reaction process strategy, the theoretical analysis and simulation of the dynamic process of global information dissemination in the online social hypernetwork are carried out. The analytical expressions between structure parameters of the hypernetwork, spreading rate and recovering rate are obtained by using the mean field theory. And then we discuss the impact of parameters including the hypernetwork scale, spreading rate, recovering rate, structure parameters of the hypernetwork as well as initial spreading nodes on the global dissemination of information. Further, a comparative analysis of the process of global information dissemination under the hypernetwork and complex network structure is carried out. The results of the study are helpful for a deeper understanding of the dissemination laws and development trends of global dissemination of information in online social network, and provide scientific basis for practical applications such as information detection and public opinion control.
In the era of information explosion, online social network has been widely depended and applied as the main way of information transmission. However, the dynamic process of information transmission is often difficult to accurately predict and prevent in online social network. In this paper, the hyperedges in hypergraphs are introduced to describe complex social relationships between two or more individuals. Based on the hypernetwork dynamic evolution model, we construct online social hypernetwork, and combine with the susceptible infected susceptible (SIS) model based on the reaction process strategy, the theoretical analysis and simulation of the dynamic process of global information dissemination in the online social hypernetwork are carried out. The analytical expressions between structure parameters of the hypernetwork, spreading rate and recovering rate are obtained by using the mean field theory. And then we discuss the impact of parameters including the hypernetwork scale, spreading rate, recovering rate, structure parameters of the hypernetwork as well as initial spreading nodes on the global dissemination of information. Further, a comparative analysis of the process of global information dissemination under the hypernetwork and complex network structure is carried out. The results of the study are helpful for a deeper understanding of the dissemination laws and development trends of global dissemination of information in online social network, and provide scientific basis for practical applications such as information detection and public opinion control.
2021, 50(3): 446-449.
doi: 10.12178/1001-0548.2021005
Abstract:
Data circulation is a novel and important means to facilitate productivity. Different from the trade of normal products, the requesting, addressing, answering and transmitting of data involve complicated procedure, and a single requirement may lead to multiple answers or even bids. Accordingly, a fundamental issue in building up an efficient and effective system for data circulation is to design a credit allocation mechanism. This paper analyzes the typical pattern of data demand and data supply, proposes an incentive network model containing requesting node, intermediary nodes and answering nodes, and designs a geometrical decaying mechanism in credit allocation. Under the above general framework, we show some typical models and the corresponding calculation processes, and extend the single chain model to the general situation involving multiple answering nodes with different weights. Lastly, we discuss how to deal with more complicated cases under this framework, such as allowing bids in competition of multiple nodes.
Data circulation is a novel and important means to facilitate productivity. Different from the trade of normal products, the requesting, addressing, answering and transmitting of data involve complicated procedure, and a single requirement may lead to multiple answers or even bids. Accordingly, a fundamental issue in building up an efficient and effective system for data circulation is to design a credit allocation mechanism. This paper analyzes the typical pattern of data demand and data supply, proposes an incentive network model containing requesting node, intermediary nodes and answering nodes, and designs a geometrical decaying mechanism in credit allocation. Under the above general framework, we show some typical models and the corresponding calculation processes, and extend the single chain model to the general situation involving multiple answering nodes with different weights. Lastly, we discuss how to deal with more complicated cases under this framework, such as allowing bids in competition of multiple nodes.
2021, 50(3): 450-458.
doi: 10.12178/1001-0548.2020338
Abstract:
In the influence maximization detection algorithm, the degree discount algorithm is an efficient heuristic algorithm. Aiming at the shortcomings of the degree discount algorithm, the formula for calculating the expected influence is modified and the first-order improved degree discount algorithm is proposed. Furthermore, in order to ensure the seed nodes are scattered in the network, a redundancy weakening mechanism is introduced and then the second-order improved degree discount algorithm is constructed. Based on the independent cascade model, the proposed algorithms are compared with other algorithms in four real networks. The experimental results confirm that the proposed algorithms can ensure faster and wider information spreading with low time complexity.
In the influence maximization detection algorithm, the degree discount algorithm is an efficient heuristic algorithm. Aiming at the shortcomings of the degree discount algorithm, the formula for calculating the expected influence is modified and the first-order improved degree discount algorithm is proposed. Furthermore, in order to ensure the seed nodes are scattered in the network, a redundancy weakening mechanism is introduced and then the second-order improved degree discount algorithm is constructed. Based on the independent cascade model, the proposed algorithms are compared with other algorithms in four real networks. The experimental results confirm that the proposed algorithms can ensure faster and wider information spreading with low time complexity.
2021, 50(3): 459-466.
doi: 10.12178/1001-0548.2020296
Abstract:
The index enhancement strategy as an organic combination of active investment and passive investment has attracted more and more attention from investors. Current index investment mainly uses methods such as machine learning to mine factors, ignoring first-hand information such as financial annual reports. This paper proposes an index enhancement strategy based on the fundamental data of accounting statements and central research of network science. First, the random forest method is used to select representative indicators in the company's accounting statements in the industry index. Second, the inter-company network is constructed based on the Pearson similarity of the indicators. Finally, the network centrality index is used to select highly central companies for portfolio investment. Research on a total of 456 stocks in 5 industry indexes shows that the return rate of the investment portfolio constructed in this article is higher and more stable than the benchmark return rate of the index. Among them, the semiconductor index's combined return rate selected in the 2019 semi-annual report is 100.37% higher than the benchmark return rate. This shows that the method has certain reference value and applicability for the research of index enhancement strategy.
The index enhancement strategy as an organic combination of active investment and passive investment has attracted more and more attention from investors. Current index investment mainly uses methods such as machine learning to mine factors, ignoring first-hand information such as financial annual reports. This paper proposes an index enhancement strategy based on the fundamental data of accounting statements and central research of network science. First, the random forest method is used to select representative indicators in the company's accounting statements in the industry index. Second, the inter-company network is constructed based on the Pearson similarity of the indicators. Finally, the network centrality index is used to select highly central companies for portfolio investment. Research on a total of 456 stocks in 5 industry indexes shows that the return rate of the investment portfolio constructed in this article is higher and more stable than the benchmark return rate of the index. Among them, the semiconductor index's combined return rate selected in the 2019 semi-annual report is 100.37% higher than the benchmark return rate. This shows that the method has certain reference value and applicability for the research of index enhancement strategy.
2021, 50(3): 467-475.
doi: 10.12178/1001-0548.2020107
Abstract:
The signal-to-noise ratio of EEG signal is low, feature extraction and feature selection are difficult, and high classification accuracy cannot be obtained. To solve these problems, this paper extracts the features of time domain, frequency domain and space domain, and uses particle swarm optimization algorithm combined with random forest classifier to screen the features. The specific process is as follows: firstly, the signal is bandpass filtered according to the R2 graph; then, the wavelet soft thresholding and scoring common space pattern algorithm are used for denoising and channel filtering; furthermore, the time-frequency domain and space domain features are extracted through three algorithms, and the evaluation index based on the random forest classifier is used as the fitness function of particle swarm optimization (PSO) after feature fusion for feature selection; finally, three classifiers and integrated classifiers are used to verify the effect. The experimental results show that through feature fusion and feature selection, redundant information can be removed and effective information can be retained. The final classification accuracy is 98.3%, which provides a new method for the application of this technology in medical rehabilitation and other fields.
The signal-to-noise ratio of EEG signal is low, feature extraction and feature selection are difficult, and high classification accuracy cannot be obtained. To solve these problems, this paper extracts the features of time domain, frequency domain and space domain, and uses particle swarm optimization algorithm combined with random forest classifier to screen the features. The specific process is as follows: firstly, the signal is bandpass filtered according to the R2 graph; then, the wavelet soft thresholding and scoring common space pattern algorithm are used for denoising and channel filtering; furthermore, the time-frequency domain and space domain features are extracted through three algorithms, and the evaluation index based on the random forest classifier is used as the fitness function of particle swarm optimization (PSO) after feature fusion for feature selection; finally, three classifiers and integrated classifiers are used to verify the effect. The experimental results show that through feature fusion and feature selection, redundant information can be removed and effective information can be retained. The final classification accuracy is 98.3%, which provides a new method for the application of this technology in medical rehabilitation and other fields.
2021, 50(3): 476-480.
doi: 10.12178/1001-0548.2021060
Abstract:
Antibody plays a key role in disease prevention, diagnosis, and treatment. Therapeutic antibodies have a market over 100 billion dollars with indications for various diseases such as cancers, autoimmune diseases, and infectious diseases. However, the trade suffers from high costs and low success rate. To assess the developability of thousands or even millions of antibody candidates comprehensively, reasonably, and quickly so as to decrease the late stage failure of antibody development is the key problem need to be solved in the field of therapeutic antibody development. This paper reviewed the advances in experimental and bioinformatics studies on evaluation of the developability of therapeutic antibodies. The problems of existing researches were also summarized. Therefore, it would benefit the relevant scholars in their further studies.
Antibody plays a key role in disease prevention, diagnosis, and treatment. Therapeutic antibodies have a market over 100 billion dollars with indications for various diseases such as cancers, autoimmune diseases, and infectious diseases. However, the trade suffers from high costs and low success rate. To assess the developability of thousands or even millions of antibody candidates comprehensively, reasonably, and quickly so as to decrease the late stage failure of antibody development is the key problem need to be solved in the field of therapeutic antibody development. This paper reviewed the advances in experimental and bioinformatics studies on evaluation of the developability of therapeutic antibodies. The problems of existing researches were also summarized. Therefore, it would benefit the relevant scholars in their further studies.
