ISSN 
2023 Vol. 52, No. 6
2023, 52(6): 802-808.
doi: 10.12178/1001-0548.2022109
Abstract:
This paper reduces the Markowitz’s model to the Quantum Linear Discriminant Analysis (QLDA) model. Hermitian Chain Product (HCP) and Density Matrix Exponentiation (DME) are used to solve the optimal solution with the largest Sharpe rate in the Markowitz mean-variance model. The quantum continuous portfolio optimization scheme can achieve quasi-exponential acceleration compared to the classical scheme.
This paper reduces the Markowitz’s model to the Quantum Linear Discriminant Analysis (QLDA) model. Hermitian Chain Product (HCP) and Density Matrix Exponentiation (DME) are used to solve the optimal solution with the largest Sharpe rate in the Markowitz mean-variance model. The quantum continuous portfolio optimization scheme can achieve quasi-exponential acceleration compared to the classical scheme.
2023, 52(6): 809-817.
doi: 10.12178/1001-0548.2022303
Abstract:
Image encryption is a key link to ensure the secure transmission of digital images over the Internet. In this paper, a color quantum image encryption algorithm based on alternating quantum random walk (AQW) and XOR operation is proposed. Firstly, the original color image is represented by color quantum image representation (NCQI), the quantum circuit is constructed, and the quantum XOR matrix is constructed by using the binary probability matrix generated by the controlled AQW to complete the encryption operation of the original quantum image, and the decryption process is the reverse process of the encryption process. The initial condition of AQW is the key to guaranteeing the security of quantum image encryption algorithms. The security of the proposed quantum image encryption model is verified through the simulation experiment of on classical computer.
Image encryption is a key link to ensure the secure transmission of digital images over the Internet. In this paper, a color quantum image encryption algorithm based on alternating quantum random walk (AQW) and XOR operation is proposed. Firstly, the original color image is represented by color quantum image representation (NCQI), the quantum circuit is constructed, and the quantum XOR matrix is constructed by using the binary probability matrix generated by the controlled AQW to complete the encryption operation of the original quantum image, and the decryption process is the reverse process of the encryption process. The initial condition of AQW is the key to guaranteeing the security of quantum image encryption algorithms. The security of the proposed quantum image encryption model is verified through the simulation experiment of on classical computer.
2023, 52(6): 818-825.
doi: 10.12178/1001-0548.2023077
Abstract:
The existing phased array equipment usually adopts sub-aperture or sub-timeslot to realize multi-target jamming, but the former cannot take advantage of full aperture gain, and the latter cannot allocate the time-domain resource smoothly when the number of targets is large. In order to solve this problem, this paper proposes a method to adjust the integer and fractional delays in order to realize the purpose of large bandwidth digital transmission. The integer delay is realized by sampling interval delay, and the fractional delay is realized by Farrow filter. The key delay characteristics are simulated, and the minimum scale digital array prototype is constructed. The test results show that the simultaneous four beams with the instantaneous bandwidth of 400 MHz are realized, the direction and power of jamming beam can be adjusted arbitrarily, and the delay characteristics are also consistent with the simulation results.
The existing phased array equipment usually adopts sub-aperture or sub-timeslot to realize multi-target jamming, but the former cannot take advantage of full aperture gain, and the latter cannot allocate the time-domain resource smoothly when the number of targets is large. In order to solve this problem, this paper proposes a method to adjust the integer and fractional delays in order to realize the purpose of large bandwidth digital transmission. The integer delay is realized by sampling interval delay, and the fractional delay is realized by Farrow filter. The key delay characteristics are simulated, and the minimum scale digital array prototype is constructed. The test results show that the simultaneous four beams with the instantaneous bandwidth of 400 MHz are realized, the direction and power of jamming beam can be adjusted arbitrarily, and the delay characteristics are also consistent with the simulation results.
2023, 52(6): 826-829.
doi: 10.12178/1001-0548.2022180
Abstract:
The conventional high frequency Log-periodic Dipole Antenna (LPDA) is very large in the lateral size, causing the construction difficulty. A miniaturized soft High Frequency (HF) LPDA work in 4 MHz to 30 MHz is presented. The elements of the antenna adopt three-dimensional triangular dipole and plane quadrilateral dipole, and the stay wires are loaded as the longest element, so that the work frequency of antenna is effectively extended. The longest antenna element adopts folded dipole antenna, which eliminates the influence flank truss on antenna performance. Full-wave simulation and test results showed a stable directional pattern and lower Voltage Standing Wave Ratio (VSWR) over the impedance.
The conventional high frequency Log-periodic Dipole Antenna (LPDA) is very large in the lateral size, causing the construction difficulty. A miniaturized soft High Frequency (HF) LPDA work in 4 MHz to 30 MHz is presented. The elements of the antenna adopt three-dimensional triangular dipole and plane quadrilateral dipole, and the stay wires are loaded as the longest element, so that the work frequency of antenna is effectively extended. The longest antenna element adopts folded dipole antenna, which eliminates the influence flank truss on antenna performance. Full-wave simulation and test results showed a stable directional pattern and lower Voltage Standing Wave Ratio (VSWR) over the impedance.
2023, 52(6): 830-834.
doi: 10.12178/1001-0548.2022227
Abstract:
Broadband infrared response has attracted great attention due to its potential applications in silicon based photodetectors. In this paper, we have fabricated a series of n+/p photodetectors with hyperdoped silicon prepared by ion-implantation and femtosecond pulsed laser. And the photoresponse spectral in near and mid-infrared region of electromagnetic spectrum of fabricated highly-doped silicon photodiodes were studied. These devices showed a remarkable photoresponse peak at near-infrared response (NIR) wavelengths. The distinct sub-band gap photoresponse features corresponding to the onset energies are consistent with the active energy levels of known sulfur within the silicon band-gap. The device fabricated with implantation dose of 1014 ions/cm2 has exhibited the best performance. This technique may offer a promising approach to fabricate low-cost broadband silicon based detectors.
Broadband infrared response has attracted great attention due to its potential applications in silicon based photodetectors. In this paper, we have fabricated a series of n+/p photodetectors with hyperdoped silicon prepared by ion-implantation and femtosecond pulsed laser. And the photoresponse spectral in near and mid-infrared region of electromagnetic spectrum of fabricated highly-doped silicon photodiodes were studied. These devices showed a remarkable photoresponse peak at near-infrared response (NIR) wavelengths. The distinct sub-band gap photoresponse features corresponding to the onset energies are consistent with the active energy levels of known sulfur within the silicon band-gap. The device fabricated with implantation dose of 1014 ions/cm2 has exhibited the best performance. This technique may offer a promising approach to fabricate low-cost broadband silicon based detectors.
2023, 52(6): 835-840.
doi: 10.12178/1001-0548.2022375
Abstract:
Power Transmission Efficiency (PTE) between transceiver antennas is one of the most significant parameters to characterize the performance of indoor near-field Microwave Power Transmission (MPT) systems. Time Reversal (TR) has been used in MPT systems as an efficient wireless power transmission technology. In this paper, the optimal solution of PTE is derived based on TR and the reciprocity theorem. At the same time, the advantages of TR over traditional MPT methods such as pattern synthesis and phase compensation methods in near-field wireless power transmission are compared based on electromagnetic full-wave simulation technology. Theory and simulation show that TR can achieve the highest PTE in the near-field region of the antenna.
Power Transmission Efficiency (PTE) between transceiver antennas is one of the most significant parameters to characterize the performance of indoor near-field Microwave Power Transmission (MPT) systems. Time Reversal (TR) has been used in MPT systems as an efficient wireless power transmission technology. In this paper, the optimal solution of PTE is derived based on TR and the reciprocity theorem. At the same time, the advantages of TR over traditional MPT methods such as pattern synthesis and phase compensation methods in near-field wireless power transmission are compared based on electromagnetic full-wave simulation technology. Theory and simulation show that TR can achieve the highest PTE in the near-field region of the antenna.
2023, 52(6): 841-850.
doi: 10.12178/1001-0548.2022253
Abstract:
Generating digital signal in complex electromagnetic environment is one of the core issues in communication countermeasures and jamming. An orthogonal frequency-division multiplexing (OFDM) signal generation scheme based on the pattern-constellation dual discriminator generative adversarial network (Pattern-Constellation dual GAN) is developed. First, symbol vectors in frequency domain are generated by applying the fast Fourier transform to the OFDM signals. Then, the symbol vectors are orderly concatenated into a two-dimensional matrix and stored as a gray-scale image, which contains the corresponding time-frequency features of the OFDM signals. Furthermore, such gray-scale images are used for training and testing the proposed Dual GAN network. In our network, an adversarial game among one generator and two discriminators is established to generate gray-scale images which contain the same time-frequency features as in the training images. The generator aims to generate a counterfeit image to confuse the two discriminators, while the two discriminators aim to distinguish the subcarrier structure and the constellation density between the generated image and the real image, respectively. Finally, the Wi-Fi 802.11a protocol signals are used as examples to verify the effectiveness of proposed signal generation model.
Generating digital signal in complex electromagnetic environment is one of the core issues in communication countermeasures and jamming. An orthogonal frequency-division multiplexing (OFDM) signal generation scheme based on the pattern-constellation dual discriminator generative adversarial network (Pattern-Constellation dual GAN) is developed. First, symbol vectors in frequency domain are generated by applying the fast Fourier transform to the OFDM signals. Then, the symbol vectors are orderly concatenated into a two-dimensional matrix and stored as a gray-scale image, which contains the corresponding time-frequency features of the OFDM signals. Furthermore, such gray-scale images are used for training and testing the proposed Dual GAN network. In our network, an adversarial game among one generator and two discriminators is established to generate gray-scale images which contain the same time-frequency features as in the training images. The generator aims to generate a counterfeit image to confuse the two discriminators, while the two discriminators aim to distinguish the subcarrier structure and the constellation density between the generated image and the real image, respectively. Finally, the Wi-Fi 802.11a protocol signals are used as examples to verify the effectiveness of proposed signal generation model.
2023, 52(6): 851-858.
doi: 10.12178/1001-0548.2022336
Abstract:
In order to improve the electromagnetic protection capability of radar equipment, the two types of pseudo-signal interference imaging mechanism of radar equipment are revealed based on the working principle of stepper-frequency radar ranging for the problem of pseudo-signal formation on the radar display interface when typical radar equipment is subjected to single-frequency electromagnetic radiation, thus causing misjudgment of real targets. With a type of Ku-band step-frequency ranging radar as the test object, a single-frequency electromagnetic radiation pseudo-signal interference effect test was carried out to verify the correctness of the two types of pseudo-signal interference imaging mechanism analysis. The variation curves of the two types of pseudo-signal level values with single-frequency interference field strength at typical frequency points in and out of the band are obtained experimentally, and the variation laws of the pseudo-signal level values in and out of the band are explained respectively based on the nonlinear distortion analysis of the receiver circuit, and then the sensitive frequency bands of the two types of pseudo-signal interference are measured. The test results show that: In-band single-frequency electromagnetic radiation can cause "pulse" type pseudo-signal interference to stepper frequency radar equipment, the location of the pseudo-signal is random; out-of-band single-frequency electromagnetic radiation can cause "impulse" type pseudo-signal interference to stepper frequency radar equipment, the location of the pseudo-signal is fixed; as the single-frequency interference field strength increases, the "pulse" type pseudo-signal level value first increases linearly, and then remains constant, the "impulse" type pseudo-signal level value first gradually increases, reaches the maximum value and then gradually decreases.
In order to improve the electromagnetic protection capability of radar equipment, the two types of pseudo-signal interference imaging mechanism of radar equipment are revealed based on the working principle of stepper-frequency radar ranging for the problem of pseudo-signal formation on the radar display interface when typical radar equipment is subjected to single-frequency electromagnetic radiation, thus causing misjudgment of real targets. With a type of Ku-band step-frequency ranging radar as the test object, a single-frequency electromagnetic radiation pseudo-signal interference effect test was carried out to verify the correctness of the two types of pseudo-signal interference imaging mechanism analysis. The variation curves of the two types of pseudo-signal level values with single-frequency interference field strength at typical frequency points in and out of the band are obtained experimentally, and the variation laws of the pseudo-signal level values in and out of the band are explained respectively based on the nonlinear distortion analysis of the receiver circuit, and then the sensitive frequency bands of the two types of pseudo-signal interference are measured. The test results show that: In-band single-frequency electromagnetic radiation can cause "pulse" type pseudo-signal interference to stepper frequency radar equipment, the location of the pseudo-signal is random; out-of-band single-frequency electromagnetic radiation can cause "impulse" type pseudo-signal interference to stepper frequency radar equipment, the location of the pseudo-signal is fixed; as the single-frequency interference field strength increases, the "pulse" type pseudo-signal level value first increases linearly, and then remains constant, the "impulse" type pseudo-signal level value first gradually increases, reaches the maximum value and then gradually decreases.
2023, 52(6): 859-865.
doi: 10.12178/1001-0548.2022263
Abstract:
There is a frequent phenomenon that optical cables pass through identical trenches in metro cable topologies. Early operations engineers used the shortest path algorithm to configure the work and protected routes of power services. With the advancement of urban infrastructure, some trenches will inevitably be damaged, leading to an increase in power grid communication service interruption numbers. This paper proposes a route optimization algorithm for power grid communication services based on minimized trenches. First, the urban optical path topology is modeled, and the output function is aimed at minimizing common trenches’ work-protect route. Then, the deep first search (DFS) fusion sorting algorithm selects all the service’s working routes. Next, the corresponding route of the work route is deleted, and all protected routes are found by using the fusion-sorted DFS algorithm again. Subsequently, the groups with the least common trench between the work and protected route are calculated and used as the final primary and secondary routes for the service. Finally, the effectiveness and practicability of the algorithm are verified by computer simulation and application examples from the Anhui metropolitan area network, respectively.
There is a frequent phenomenon that optical cables pass through identical trenches in metro cable topologies. Early operations engineers used the shortest path algorithm to configure the work and protected routes of power services. With the advancement of urban infrastructure, some trenches will inevitably be damaged, leading to an increase in power grid communication service interruption numbers. This paper proposes a route optimization algorithm for power grid communication services based on minimized trenches. First, the urban optical path topology is modeled, and the output function is aimed at minimizing common trenches’ work-protect route. Then, the deep first search (DFS) fusion sorting algorithm selects all the service’s working routes. Next, the corresponding route of the work route is deleted, and all protected routes are found by using the fusion-sorted DFS algorithm again. Subsequently, the groups with the least common trench between the work and protected route are calculated and used as the final primary and secondary routes for the service. Finally, the effectiveness and practicability of the algorithm are verified by computer simulation and application examples from the Anhui metropolitan area network, respectively.
2023, 52(6): 866-871.
doi: 10.12178/1001-0548.2022361
Abstract:
In this paper, a study is conducted on corn haploid seeds recognition based on convolutional neural network using 3000 corn seed images with 1230 haploid corn seed images and 1770 diploid corn seed images. In order to compare the effect of different convolutional neural network models on haploid corn seeds recognition, classical models including VGG, ResNet, DenseNet and SKNet are adopted, and the SKNet model is improved by replacing the fully-connected layer in dimensionality reduction and dimensionality increase with one-dimensional convolution to further reduce the number of model parameters, and the improved SKNet is called ECA_SKNet. The experimental results show that aforementioned five models can achieve good recognition of haploid corn seeds with the lowest accuracy of 88.5% and the accuracy of ECA_SKNet can reach 93.04%. It is seen that convolutional neural networks can play an important role for the recognition of corn haploid seeds and provide a new way to recognize crop seeds.
In this paper, a study is conducted on corn haploid seeds recognition based on convolutional neural network using 3000 corn seed images with 1230 haploid corn seed images and 1770 diploid corn seed images. In order to compare the effect of different convolutional neural network models on haploid corn seeds recognition, classical models including VGG, ResNet, DenseNet and SKNet are adopted, and the SKNet model is improved by replacing the fully-connected layer in dimensionality reduction and dimensionality increase with one-dimensional convolution to further reduce the number of model parameters, and the improved SKNet is called ECA_SKNet. The experimental results show that aforementioned five models can achieve good recognition of haploid corn seeds with the lowest accuracy of 88.5% and the accuracy of ECA_SKNet can reach 93.04%. It is seen that convolutional neural networks can play an important role for the recognition of corn haploid seeds and provide a new way to recognize crop seeds.
Dynamic Prior Features-Based Deep Learning Framework for Multidefect Detection of Coated Propellants
2023, 52(6): 872-879.
doi: 10.12178/1001-0548.2022326
Abstract:
Coated propellants (CPs) are extensively used in the dynamical systems of rockets and missiles. The appearance quality of the CPs has significant impact on the performance of the systems. To this end, a dynamic prior features-based deep learning framework for multidefect detection of CPs, such as shape, size, and surface defects, is put forth in this article: 1) An integrated deep model for deep classifier (DC)-based shape defect and deep segmentation network (DSN)-based size defect detection is introduced, which can remove redundant features among different tasks. Particularly, the features generated by the current iteration of the DSN, as dynamic prior features, act on the next iteration of the DC to accelerate the convergence rate, and 2) the dynamic features are also mapped to the convolutional autoencoder-based surface defect detection, which can guide the model to quickly focus on the CPs, while suppressing the repeated feature extraction of task-independent features. Experimental results on an image dataset from a real-world manufacturing line show that the proposed framework has the superiority in terms of the power consumption, detection efficiency, and detection accuracy.
Coated propellants (CPs) are extensively used in the dynamical systems of rockets and missiles. The appearance quality of the CPs has significant impact on the performance of the systems. To this end, a dynamic prior features-based deep learning framework for multidefect detection of CPs, such as shape, size, and surface defects, is put forth in this article: 1) An integrated deep model for deep classifier (DC)-based shape defect and deep segmentation network (DSN)-based size defect detection is introduced, which can remove redundant features among different tasks. Particularly, the features generated by the current iteration of the DSN, as dynamic prior features, act on the next iteration of the DC to accelerate the convergence rate, and 2) the dynamic features are also mapped to the convolutional autoencoder-based surface defect detection, which can guide the model to quickly focus on the CPs, while suppressing the repeated feature extraction of task-independent features. Experimental results on an image dataset from a real-world manufacturing line show that the proposed framework has the superiority in terms of the power consumption, detection efficiency, and detection accuracy.
2023, 52(6): 880-886.
doi: 10.12178/1001-0548.2022363
Abstract:
Aiming at the difficulty of deep learning model training caused by the lack of labeled lung Computed Tomography (CT) image data and the lung nodule feature model generated by existing generation algorithmsTo solve the problem of blur and detail loss, a data-enhanced RAU-GAN algorithm for pulmonary nodule images is proposed. Firstly, a residual attention module is embedded in the generator network, which can focus on different local regions of interest to achieve the independent generation of lung nodules and background information. Moreover, the residual block structure in the attention module is redesigned to to reduce the depth of the network and training complexity. Second, the discriminator is designed as U-Net architecture, which can feed back more information to the updated generator to improve the discrimination performance. Finally, experiments were conducted on data set LUNA16 and deep lesion. The results show that the visual and different evaluation indexes have improved in comparison with existing methods, which verifies that the generated images can contain richer details. images can contain richer details.
Aiming at the difficulty of deep learning model training caused by the lack of labeled lung Computed Tomography (CT) image data and the lung nodule feature model generated by existing generation algorithmsTo solve the problem of blur and detail loss, a data-enhanced RAU-GAN algorithm for pulmonary nodule images is proposed. Firstly, a residual attention module is embedded in the generator network, which can focus on different local regions of interest to achieve the independent generation of lung nodules and background information. Moreover, the residual block structure in the attention module is redesigned to to reduce the depth of the network and training complexity. Second, the discriminator is designed as U-Net architecture, which can feed back more information to the updated generator to improve the discrimination performance. Finally, experiments were conducted on data set LUNA16 and deep lesion. The results show that the visual and different evaluation indexes have improved in comparison with existing methods, which verifies that the generated images can contain richer details. images can contain richer details.
2023, 52(6): 887-896.
doi: 10.12178/1001-0548.2022278
Abstract:
The current Shapelet extraction algorithm for time series ordinal classification, which suffers from low efficiency, needs to figure out the Pearson's correlation coefficient or the Spearman's correlation coefficient between the Euclidean distances and the label distances from time series to Shapelets to evaluate the Shapelets. To handle this problem, this paper first proposes a Shapelet measure CD-Cover (concentration and dominance of coverage) based on the SAX (symbolic aggregate approximation)-represented time series. The measure takes into account both the concentration and the dominance of coverage of a Shapelet on the time series dataset. Secondly, this paper also proposes a Shapelet extraction algorithm based on random sampling. The algorithm uses the Bloom filter to pre-prune Shapelet candidates and employs a strategy of removing self-similar Shapelets to post-prune the extracting results. Experimental results on 11 time series public datasets show that the Shapelet extracted by the proposed algorithm has better ability for ordinal classification than the existing methods, and meanwhile, the computing efficiency of the proposed algorithm is superior to that of the existing methods.
The current Shapelet extraction algorithm for time series ordinal classification, which suffers from low efficiency, needs to figure out the Pearson's correlation coefficient or the Spearman's correlation coefficient between the Euclidean distances and the label distances from time series to Shapelets to evaluate the Shapelets. To handle this problem, this paper first proposes a Shapelet measure CD-Cover (concentration and dominance of coverage) based on the SAX (symbolic aggregate approximation)-represented time series. The measure takes into account both the concentration and the dominance of coverage of a Shapelet on the time series dataset. Secondly, this paper also proposes a Shapelet extraction algorithm based on random sampling. The algorithm uses the Bloom filter to pre-prune Shapelet candidates and employs a strategy of removing self-similar Shapelets to post-prune the extracting results. Experimental results on 11 time series public datasets show that the Shapelet extracted by the proposed algorithm has better ability for ordinal classification than the existing methods, and meanwhile, the computing efficiency of the proposed algorithm is superior to that of the existing methods.
2023, 52(6): 897-905.
doi: 10.12178/1001-0548.2022407
Abstract:
Aiming at the problems of data theft and privacy leakage in the process of digital image calculation, storage and transmission, a multi-image encryption method based on hyperchaotic system with variable parameters is proposed. Firstly, the state variables of one chaotic system are used to perturb the state parameters of the other chaotic system, and a hyperchaotic system with variable parameters is constructed. Secondly, the input gray image pair is reconstructed and input into SHA-512 algorithm to generate the initial key. Then, the initial key is input into the hyperchaotic system with variable parameters to generate five groups of chaotic sequences iteratively, and then the reconstructed image is transformed into a magic square to achieve the transformation of pixel positions. Finally, the S-shaped diffusion is performed on the image obtained by the magic square transformation to realize the change of the pixel value, and an approximately uniformly distributed ciphertext image is obtained. The results show that the algorithm improves the low randomness and low complexity of the traditional image encryption method, and at the same time, and improves the disorder of the ciphertext image and the ability of the encryption method to resist conventional attacks.
Aiming at the problems of data theft and privacy leakage in the process of digital image calculation, storage and transmission, a multi-image encryption method based on hyperchaotic system with variable parameters is proposed. Firstly, the state variables of one chaotic system are used to perturb the state parameters of the other chaotic system, and a hyperchaotic system with variable parameters is constructed. Secondly, the input gray image pair is reconstructed and input into SHA-512 algorithm to generate the initial key. Then, the initial key is input into the hyperchaotic system with variable parameters to generate five groups of chaotic sequences iteratively, and then the reconstructed image is transformed into a magic square to achieve the transformation of pixel positions. Finally, the S-shaped diffusion is performed on the image obtained by the magic square transformation to realize the change of the pixel value, and an approximately uniformly distributed ciphertext image is obtained. The results show that the algorithm improves the low randomness and low complexity of the traditional image encryption method, and at the same time, and improves the disorder of the ciphertext image and the ability of the encryption method to resist conventional attacks.
2023, 52(6): 906-914.
doi: 10.12178/1001-0548.2022290
Abstract:
In the defect detection of transmission lines, the bird's nest, plastic, rags and other suspended solids are mostly small targets. They have few pixels in the images and are easy to be disturbed by the background, which make the detection accuracy needs to be improved. In this paper, a new two-stage object detection algorithm is designed to improve the detection effect of bird nests and suspended solids in transmission lines. In order to improve the detection performance of small targets, the attention mechanism is integrated into the feature extraction network to learn more rich context information. In addition, in the detection network, a post-processing method based on softer non maximum suppression algorithm is designed to reduce the loss of small targets. Compared with the commonly used two-stage object detection algorithms, the proposed method improves the average accuracy of the two categories by about 4.7% and 5.9%, respectively, and has greater value in practical applications.
In the defect detection of transmission lines, the bird's nest, plastic, rags and other suspended solids are mostly small targets. They have few pixels in the images and are easy to be disturbed by the background, which make the detection accuracy needs to be improved. In this paper, a new two-stage object detection algorithm is designed to improve the detection effect of bird nests and suspended solids in transmission lines. In order to improve the detection performance of small targets, the attention mechanism is integrated into the feature extraction network to learn more rich context information. In addition, in the detection network, a post-processing method based on softer non maximum suppression algorithm is designed to reduce the loss of small targets. Compared with the commonly used two-stage object detection algorithms, the proposed method improves the average accuracy of the two categories by about 4.7% and 5.9%, respectively, and has greater value in practical applications.
2023, 52(6): 915-924.
doi: 10.12178/1001-0548.2022225
Abstract:
Mobile crowdsensing technology is a technique that can break through the limitations of time and place, and realize large-scale real-time crowdsensing data perception, transmission and sharing anytime, anywhere. However, the existing mobile crowdsensing applications confront with some security, privacy and efficiency problems in the crowdsensing data sharing, such as unauthorized data access, privacy leakage of access control, key escrow of single-authority, and high access overhead. In order to tackle the above problems simultaneously, this paper proposes an attribute-hiding based efficient and decentralized scheme for mobile crowdsensing data sharing, which not only allows mobile users to specify attribute-based access control for encrypting crowdsensing data, such that only users who meet the access control can access the data, but also allows multiple authorities to jointly generate private keys for swarm crowdsensing users, so that any a single authority cannot illegally access the target crowdsensing data by pretending to be a legitimate user. In addition, it enables fast decryption and accessing target data with the lowest energy consumption without leaking attribute privacy of access control. This paper also gives strict security analysis and performance analysis to prove that our scheme is secure, efficient and feasible for the mobile crowdsensing data sharing.
Mobile crowdsensing technology is a technique that can break through the limitations of time and place, and realize large-scale real-time crowdsensing data perception, transmission and sharing anytime, anywhere. However, the existing mobile crowdsensing applications confront with some security, privacy and efficiency problems in the crowdsensing data sharing, such as unauthorized data access, privacy leakage of access control, key escrow of single-authority, and high access overhead. In order to tackle the above problems simultaneously, this paper proposes an attribute-hiding based efficient and decentralized scheme for mobile crowdsensing data sharing, which not only allows mobile users to specify attribute-based access control for encrypting crowdsensing data, such that only users who meet the access control can access the data, but also allows multiple authorities to jointly generate private keys for swarm crowdsensing users, so that any a single authority cannot illegally access the target crowdsensing data by pretending to be a legitimate user. In addition, it enables fast decryption and accessing target data with the lowest energy consumption without leaking attribute privacy of access control. This paper also gives strict security analysis and performance analysis to prove that our scheme is secure, efficient and feasible for the mobile crowdsensing data sharing.
2023, 52(6): 925-931.
doi: 10.12178/1001-0548.2023108
Abstract:
The detection and recognition of helicopter rotor acoustic signals are the important problems in the field of low altitude target warning. The current algorithms treated detection and recognition separately, but in practical applications, detection and recognition is a whole process. In order to solve the above problems, this paper proposes an integrated detection and recognition algorithm for helicopter rotor acoustic signal based on deep learning. Firstly, the target is detected by fusing feature extraction and support vector machine, and then the signal segments of potential helicopter acoustic targets are classified and recognized based on deep learning. The effect of the algorithm is studied and tested in detail by constructing experimental data. Experimental results show that the detection rate and recognition rate of the algorithm are greatly improved, and the detection accuracy and recognition accuracy are greatly enhanced. The algorithm has high research and application values.
The detection and recognition of helicopter rotor acoustic signals are the important problems in the field of low altitude target warning. The current algorithms treated detection and recognition separately, but in practical applications, detection and recognition is a whole process. In order to solve the above problems, this paper proposes an integrated detection and recognition algorithm for helicopter rotor acoustic signal based on deep learning. Firstly, the target is detected by fusing feature extraction and support vector machine, and then the signal segments of potential helicopter acoustic targets are classified and recognized based on deep learning. The effect of the algorithm is studied and tested in detail by constructing experimental data. Experimental results show that the detection rate and recognition rate of the algorithm are greatly improved, and the detection accuracy and recognition accuracy are greatly enhanced. The algorithm has high research and application values.
2023, 52(6): 932-943.
doi: 10.12178/1001-0548.2022289
Abstract:
In recent years, object recognition modeling techniques based on deep learning face new challenges such as insufficient annotated samples, low interpretability of models, and insufficient stability. All these challenges limit the possibility of deep learning to solve more complex and abstract problems. Constructing intelligent model jointly driven by knowledge and data is an important way to break through the existing bottleneck. This paper presents a classification standard of model constructing methods according to the way of introduction of external experience and cognitive knowledge during model constructing, including modeling methods based on explicit knowledge, modeling methods based on implicit knowledge and modeling methods based on fusion knowledge. Then, following the proposed classification standard, the explorations in each class of methods about solving the problems of few samples and model interpretability are reviewed. Subsequently, taking advantage of different model constructing methods, a future model constructing method jointly driven by knowledge and data is proposed. The proposed method can effectively solve the problem of model construction under the condition of few samples and improve the interpretability of the model by decoupling knowledge modeling and data modeling and taking unsupervised and weak supervised training as the core training patterns. Finally, some research issues which need further study as well as future research directions are drawn in conclusion for promote the object recognition model constructing.
In recent years, object recognition modeling techniques based on deep learning face new challenges such as insufficient annotated samples, low interpretability of models, and insufficient stability. All these challenges limit the possibility of deep learning to solve more complex and abstract problems. Constructing intelligent model jointly driven by knowledge and data is an important way to break through the existing bottleneck. This paper presents a classification standard of model constructing methods according to the way of introduction of external experience and cognitive knowledge during model constructing, including modeling methods based on explicit knowledge, modeling methods based on implicit knowledge and modeling methods based on fusion knowledge. Then, following the proposed classification standard, the explorations in each class of methods about solving the problems of few samples and model interpretability are reviewed. Subsequently, taking advantage of different model constructing methods, a future model constructing method jointly driven by knowledge and data is proposed. The proposed method can effectively solve the problem of model construction under the condition of few samples and improve the interpretability of the model by decoupling knowledge modeling and data modeling and taking unsupervised and weak supervised training as the core training patterns. Finally, some research issues which need further study as well as future research directions are drawn in conclusion for promote the object recognition model constructing.
2023, 52(6): 944-953.
doi: 10.12178/1001-0548.2023058
Abstract:
How to efficiently measure the importance of nodes has been a hot issue in the research of complex networks. In the research of node importance, many algorithms have been proposed to judge key nodes, but most of them are limited to high time complexity or single evaluation angle. Considering that entropy can be used to quantitatively describe the amount of information, this paper proposes a node importance ranking algorithm based on cross entropy. This algorithm takes into account the overall influence among the central node and its neighbor nodes, organically fuses the neighborhood topology information of nodes, and uses cross entropy to quantify the information differences between nodes. In order to verify the performance of the algorithm, this paper first uses monotone relation, maximum connectivity coefficient, network efficiency and SIR model as evaluation indicators, and then compares with other seven algorithms on eight real networks in different fields. The experimental results show that the algorithm proposed in this paper is effective and applicable, and the time complexity is only\begin{document}$ O(n) $\end{document} ![]()
![]()
How to efficiently measure the importance of nodes has been a hot issue in the research of complex networks. In the research of node importance, many algorithms have been proposed to judge key nodes, but most of them are limited to high time complexity or single evaluation angle. Considering that entropy can be used to quantitatively describe the amount of information, this paper proposes a node importance ranking algorithm based on cross entropy. This algorithm takes into account the overall influence among the central node and its neighbor nodes, organically fuses the neighborhood topology information of nodes, and uses cross entropy to quantify the information differences between nodes. In order to verify the performance of the algorithm, this paper first uses monotone relation, maximum connectivity coefficient, network efficiency and SIR model as evaluation indicators, and then compares with other seven algorithms on eight real networks in different fields. The experimental results show that the algorithm proposed in this paper is effective and applicable, and the time complexity is only
2023, 52(6): 954-960.
doi: 10.12178/1001-0548.2022235
Abstract:
The existing researches lack a uniform and valid paradigm for qualitative and quantitative analysis of the influence of network microstructure on community characteristics. In addition, the existing researches are not convincing due to limited experimental data set and insufficient consideration of the coupling between various factors. For the purpose of qualitative analysis, we conduct the analysis of the influence of microstructure on community characteristics based on null models and “Significance Test”, and complete the community structure significance detection for various types of networks, thereby accomplishing the qualitative analysis of the influence of microstructure on community characteristics. To quantitatively analyze how microstructure affects community characteristic, the paper proposes a model based on null models and “Mediating Effects Analysis”. Additionally, it also analyzes the differences of modularity values between the original network and null models or between null models and null models of community structure significance type network, eliminates the effects of microstructure on community characteristics, and quantifies the contributions of various types of network microstructure to community characteristics of different community structure significance types of networks. A comprehensive analysis of the function of microstructure on the characteristics of communities is conducted based on 550 empirical networks of different scales, including social biology, science and technology, transportation, economy, and information. This will enable a deeper understanding of the mechanism responsible for the formation of community characteristics.
The existing researches lack a uniform and valid paradigm for qualitative and quantitative analysis of the influence of network microstructure on community characteristics. In addition, the existing researches are not convincing due to limited experimental data set and insufficient consideration of the coupling between various factors. For the purpose of qualitative analysis, we conduct the analysis of the influence of microstructure on community characteristics based on null models and “Significance Test”, and complete the community structure significance detection for various types of networks, thereby accomplishing the qualitative analysis of the influence of microstructure on community characteristics. To quantitatively analyze how microstructure affects community characteristic, the paper proposes a model based on null models and “Mediating Effects Analysis”. Additionally, it also analyzes the differences of modularity values between the original network and null models or between null models and null models of community structure significance type network, eliminates the effects of microstructure on community characteristics, and quantifies the contributions of various types of network microstructure to community characteristics of different community structure significance types of networks. A comprehensive analysis of the function of microstructure on the characteristics of communities is conducted based on 550 empirical networks of different scales, including social biology, science and technology, transportation, economy, and information. This will enable a deeper understanding of the mechanism responsible for the formation of community characteristics.
