本期目录

2018年 25卷 6期 刊出日期：2018-12-30



Artificial Intelligence

Pulmonary tuberculosis detection model of chest X-ray images using convolutional neural network

He Jin, Wang Cong, Chen Zhao

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1022

摘要 ( 519 ) PDF (0 KB)( 231 )

参考文献 | 相关文章 | 计量指标

The primary screening for pulmonary tuberculosis mainly relies on X-ray imaging all over the world. In recent years, the incidence of pulmonary tuberculosis has rebounded. This paper proposes a convolutional neural networks (CNN) based model on the tuberculosis detection of chest X-ray images, which is used for the automatic screening of pulmonary tuberculosis. Compared with the conventional CNN, this model can be used to detect the details of images and the areas of the disease quickly and accurately. There is an improvement in the learning speed and accuracy rate of our method, so it can better complete the work of anomaly detection and it can provide more effective auxiliary decision information for the practitioners.

Consortium blockchains-based deep deterministic policy gradient algorithm for optimal electricity trading among households

Yang Chen

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1023

摘要 ( 629 ) PDF (0 KB)( 142 )

参考文献 | 相关文章 | 计量指标

To achieve higher energy utilization and lower generation cost for renewable sources ( e. g. , wind and solar energy), much work has been focused on demand response in smart grid (SG). Nonetheless, most existing studies consider energy trading with utility company which results in high energy loss from high voltage to low voltage and privacy leakage. Besides, there are relatively few researches devoted to electricity scheduling and price optimum among households without a third party. To cope with these issues, a novel deep deterministic policy gradient (DDPG)-based energy trading method with consortium blockchain (DETCB) is introduced. Firstly, in order to enhance system security, executing energy transaction among households is on the basis of consortium blockchain, which leads to not only anonymous trade but also public account. Moreover, the primary target from the aspect of the system is apparently the maximal social welfare, thus exploiting an iterative decision-making method combined with DDPG algorithm by non-profit controllers to obtain optimal trading prices and carry out optimal electricity allocation. To this end, security analysis demonstrates that DETCB contributes to creating a secure, stable and trustful environment. Furthermore, the excellent performance concerning social welfare, algorithm efficiency, and transaction energy sum is shown by numerical results.

Autonomous driving in the uncertain traffic -- a deep reinforcement learning approach

Yang Shun, Wu Jian, Zhang Sumin, Han Wei

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1024

摘要 ( 519 ) PDF (0 KB)( 137 )

参考文献 | 相关文章 | 计量指标

Driving in the complex traffic safely and efficiently is a difficult task for autonomous vehicle because of the stochastic characteristics of engaged human drivers. Deep reinforcement learning (DRL), which combines the abstract representation capability of deep learning (DL) and the optimal decision making and control capability of reinforcement learning (RL), is a good approach to address this problem. Traffic environment is built up by combining intelligent driver model (IDM) and lane-change model as behavioral model for vehicles. To increase the stochastic of the established traffic environment, tricks such as defining a speed distribution with cutoff for traffic cars and using various politeness factors to represent distinguished lane-change style, are taken. For training an
artificial agent to achieve successful strategies that lead to the greatest long-term rewards and sophisticated maneuver, deep deterministic policy gradient (DDPG) algorithm is deployed for learning. Reward function is designed to get a trade-off between the vehicle speed, stability and driving safety. Results show that the proposed approach can achieve good autonomous maneuvering in a scenario of complex traffic behavior through interaction with the environment.

Pedestrian trajectory prediction in crossing scenario using fuzzy logic and switching Kalman filter

Wang Likun, Liu Lu, Yu Dameng, Xu Qing, Wang Jianqiang

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1025

摘要 ( 529 ) PDF (0 KB)( 207 )

参考文献 | 相关文章 | 计量指标

Pedestrian trajectory prediction plays an important role in bothadvanced driving assistance system (ADAS) and autonomous vehicles. An algorithm for pedestrian trajectory prediction in crossing scenario is proposed. To obtain features of pedestrian motion, we develop a method for data labelling and pedestrian body orientation regression. Using the hierarchical features as domain of discourse, fuzzy logic rules are built to describe the transition between
different pedestrian states and motion models. With derived probability of each type of motion model we further predict the pedestrian trajectory in the next 1.5 s using switching Kalman filter (KF). The proposed algorithm is further verified in our dataset, and the result indicates that the proposed algorithm successfully predicts pedestrian' s crossing behavior 0.4 s earlier before pedestrian moves. Meanwhile, the precision of predicted trajectory surpasses
other methods including interacting multi-model KF and dynamic Bayesian network (DBN).

Novel wolf pack optimization algorithm for intelligent medical treatment personalized recommendation

Wang Dongxing, Qian Xu, Liu Kang, Guan Xinyu

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1026

摘要 ( 447 ) PDF (0 KB)( 162 )

参考文献 | 相关文章 | 计量指标

To help the people choose a proper medical treatment organizer, this paper proposes an opposition raiding wolf pack optimization algorithm using random search strategy ( ORRSS-WPOA) for an adaptive shrinking region. Firstly, via the oppositional raiding method (ORM), each wolf has bigger probability of approaching the leader wolf, which makes the exploration of the wolf pack enhanced as a whole. In another word, the wolf pack is not easy to fall into local optimum. Moreover, random searching strategy (RSS) for an adaptive shrinking region is adopted to strengthen exploitation, which enables any wolf to be more likely to find the optimum in some a given region, so macroscopically the wolf pack is easier to find the global optimal in the given range. Finally, a fitness function was designed to judge the appropriateness between a certain patient and a hospital. The performance of the ORRSS-WPOA was comprehensively evaluated by comparing it with several other competitive algorithms on ten classical benchmark functions and the simulated fitness function aimed to solve the problem mentioned above. Under the same condition, our experimental results indicated the excellent performance of ORRSS-WPOA in terms of solution quality and computational efficiency.

Steering control in autonomous vehicles using deep reinforcement learning

Xue Chong, Jia Peng, Zhang Xinyu

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1027

摘要 ( 700 ) PDF (0 KB)( 181 )

参考文献 | 相关文章 | 计量指标

A novel deep reinforcement learning-based steering control method of autonomous vehicles is proposed. A distortionless compressing method of action space is presented. Convolutional neural networks (CNNs) are designed to serve as an action policy. Driver experience is investigated and modeled to optimize policy of new actions exploration. Experimental results show that the proposed algorithm has better robustness and smoothness. Moreover, it is applicable to different roads, velocities or wire-control systems.

Comparison of three data mining methods in predicting 5-year survival of colorectal cancer patients

Luo Yan, Sun Yawei, Fu Qunchao, Xue Tengfei, Zhou Ping

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1028

摘要 ( 470 ) PDF (0 KB)( 114 )

参考文献 | 相关文章 | 计量指标

The prediction of colorectal cancer (CRC) survivability has always been a challenging research issue. Considering the importance of predicting CRC patients' survival rates, we compared the performance of three data mining methods: decision trees (DTs), artificial neural networks (ANNs) and support vector machines (SVMs), for predicting 5-year survival of CRC patients to assist clinicians in making treatment decisions. The CRC dataset used to build the prediction model comes from the surveillance, epidemiology, and end results (SEER) program. The 5-fold cross-validation and random forest algorithm were respectively utilized for measuring the model predictive accuracy and the importance of features. Experimental results show that the predictive accuracy of ANNs (0.73) and SVMs (0.75) were higher than that of DTs, and they also have the best result in the area under the receiver operating characteristic (ROC) curve (area under curve (AUC) =0.82). This result may indicate high predictive power of ANNs and SVMs for predicting 5-year survival of CRC patients.

Signal processing

Construction of compressed sensing matrices based on affine symplectic space over finite fields

Wang Gang, Niu Minyao, Fu Fangwei

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1029

摘要 ( 419 ) PDF (0 KB)( 139 )

参考文献 | 相关文章 | 计量指标

The compressed sensing matrices based on affine symplectic space are constructed. Meanwhile, a comparison is made with the compressed sensing matrices constructed by DeVore based on polynomials over finite fields. Moreover, we merge our binary matrices with other low coherence matrices such as Hadamard matrices and discrete fourier transform (DFT) matrices using the embedding operation. In the numerical simulations, our matrices and modified matrices are superior to Gaussian matrices and DeVore's matrices in the performance of recovering original signals.

Computationally efficient 2-D DOA estimation for non-uniform two-L-shaped array

Yang Jinsheng, Xiang Yang, Chen Weigang, Dong Yangyang

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1030

摘要 ( 390 ) PDF (0 KB)( 140 )

参考文献 | 相关文章 | 计量指标

A two-dimensional direction-of-arrival (DOA) estimation method for non-uniform two-L-shaped array is presented in which the element spacing is larger than half-wavelength. To extract automatically paired low-variance cyclically ambiguous direction cosines and high-variance unambiguous direction cosines from the sub-blocks, the proposed method constructs and partitions the cross-correlation matrices. Then, the low-variance unambiguous direction cosines are obtained using the ambiguity resolved technique. Simulation results demonstrate that the proposed method has lower computation complexity and higher resolution than the existing methods especially when the elevation angles are between 70 and 90 degrees.

Wireless

Modified min-sum SCAN decoding algorithm for polar codes

Yuan Fang, Tian Bin

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1031

摘要 ( 480 ) PDF (0 KB)( 149 )

参考文献 | 相关文章 | 计量指标

The state-of-the-art soft-output decoder of polar codes is the soft cancellation (SCAN) decoding algorithm, which performs well at the cost of plentiful computations. Based on the SCAN decoding algorithm, a modified method with revised iterative formula is proposed, marked modified min-sum SCAN (MMS-SCAN). The proposed algorithm simplifies the update formula of nodes and reduces the complexity of iterative decoding process by the piecewise
approximation function. Meanwhile, the bit error rate (BER) of the proposed method can approach the performance of original SCAN decoding method without performance loss. The simulation reveals that the MMS-SCAN decoding algorithm can achieve the effect that the BER curve almost coincides with the original SCAN decoding curve.

Low complexity detection algorithm based on optimized Neumann series for massive MIMO system

Qu Tuosi, Cao Haiyan, Xu Fangmin, Wang Xiumin

中国邮电高校学报(英文), 2018, 25 (6). doi： 10.19682/j.cnki.1005-8885.2018.1032

摘要 ( 399 ) PDF (0 KB)( 116 )

参考文献 | 相关文章 | 计量指标

An optimized Neumann series ( NS ) approximation is described based on Frobenius matrix decomposition, this method aims to reduce the high complexity, which caused by the large matrix inversion of detection algorithm in the massive multiple input multiple output (MIMO) system. The large matrix in the inversion is decomposed into the sum of the hollow matrix and a Frobenius matrix, and the Frobenius matrix has the diagonal elements and the first column of the large matrix. In order to ensure the detection performance approach to minimum mean square error (MMSE) algorithm, the first three terms of the series approximation are needed, which results in high complexity as O(K³), where K is the number of users. This paper further optimize the third term of the series approximation to reduce the computational complexity from O(K³) to O(K²). The computational complexity analysis and simulation results show that the performance of proposed algorithm can approach to MMSE algorithm with low complexity O(K²).