本期目录

2020年 27卷 5期 刊出日期：2020-10-22



Artificial Intelligence

Collaborative filtering recommendation algorithm based on interactive data classification

季一木, 李可, 刘尚东, 刘强, 尧海昌, 李奎

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0024

摘要 ( 853 ) PDF (1589 KB)( 244 )

相关文章 | 计量指标

In the matrix factorization (MF) based collaborative filtering recommendation method, the most critical part is to deal with the interaction between the features of users and items. The mainstream approach is to use the inner product for MF to describe the user-item relationship. However, as a shallow model, MF has its limitations in describing the relationship between data. In addition, when the size of the data is large, the performance of MF is often poor due to data sparsity and noise. This paper presents a model called PIDC, short for potential interaction data clustering based deep learning recommendation. First, it uses classifiers to filter and cluster recommended items to solve the problem of sparse training data. Second, it combines MF and multi-layer perceptron (MLP) to optimize the prediction effect, and the limitation of inner product on the model expression ability is eliminated. The proposed model PIDC is tested on two datasets. The experimental results show that compared with the existing benchmark algorithm, the model improved the recommendation effect.

Fast TMRM: efficient multi-task recommendation model

朱帆, 杨娟

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0025

摘要 ( 551 ) PDF (2000 KB)( 96 )

参考文献 | 相关文章 | 计量指标

An improved multi-task learning recommendation algorithm—fast two-stage multi-task recommendation model boosted feature selection (Fast TMRM) is proposed based on auto-encoders in this paper. Compared to previous work, Fast TMRM improves the convergence speed and accuracy of training. In addition, Fast TMRM builds on previous work to introduce the auto-encoder to encode the important feature combination vector. That is how it can be used for the training of multi-task learning, which helps to improve the training efficiency of the model by nearly 67%. Finally, the nearest neighbor search is used to restore important feature expression.


Semantic segmentation of track image based on deep neural network

汪昭颖, 周军华, 廖中华, 翟翔, 张连平

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0023

摘要 ( 581 ) PDF (1624 KB)( 204 )

参考文献 | 相关文章 | 计量指标

In this paper, deep learning technology was utilited to solve the railway track recognition in intrusion detection problem. The railway track recognition can be viewed as semantic segmentation task which extends image processing to pixel level prediction. An encoder-decoder architecture DeepLabv3 + model was applied in this work due to its good performance in semantic segmentation task. Since images of the railway track collected from the video surveillance of the train cab were used as experiment dataset in this work, the following improvements were made to the model. The first aspect deals with over-fitting problem due to the limited amount of training data. Data augmentation and transfer learning are applied consequently to rich the diversity of data and enhance model robustness during the training process. Besides, different gradient descent methods are compared to obtain the optimal optimizer for training model parameters. The third problem relates to data sample imbalance, cross entropy (CE) loss is replaced by focal loss (FL) to address the issue of serious imbalance between positive and negative sample. Effectiveness of the improved DeepLabv3 + model with above solutions is demonstrated by experiment results with different system parameters.


Web log classification framework with data augmentation based on GANs

何明枢, 金磊, 王小娟, 李源

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0020

摘要 ( 600 ) PDF (1352 KB)( 695 )

参考文献 | 相关文章 | 计量指标

Attacks on web servers are part of the most serious threats in network security fields. Analyzing logs of web attacks is an effective approach for malicious behavior identification. Traditionally, machine learning models based on labeled data are popular identification methods. Some deep learning models are also recently introduced for analyzing logs based on web logs classification. However, it is limited to the amount of labeled data in model training. Web logs with labels which mark specific categories of data are difficult to obtain. Consequently, it is necessary to follow the problem about data generation with a focus on learning similar feature representations from the original data and improve the accuracy of classification model. In this paper, a novel framework is proposed, which differs in two important aspects: one is that long short-term memory (LSTM) is incorporated into generative adversarial networks (GANs) to generate the logs of web attack. The other is that a data augment model is proposed by adding logs of web attack generated by GANs to the original dataset and improved the performance of the classification model. The results experimentally demonstrate the effectiveness of the proposed method. It improved the classification accuracy from 89.04% to 95.04%.


Wireless

MIMO-FSK non-coherent detection with spatial multiplexing in fast-fading environment

郑霖, 汪震, 陈建梅, 林梦莹, 邓小芳

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0022

摘要 ( 511 ) PDF (5190 KB)( 163 )

参考文献 | 相关文章 | 计量指标

Accurate estimationand real-time compensation for phase offset and Doppler shift are essential for coherent multi-input multi-output (MIMO) systems. Here, a spatial multiplexing MIMO scheme with non-coherent frequency-shift keying (FSK) detection is proposed. It is immune to random phase interference and Doppler shift while increasing capacity. It is valuable that the proposed spatial multiplexing MIMO based on energy detection (ED) is equivalent to a linear system, and there is no mutual interference caused by the product of simultaneous signals in square-law processing. The equivalent MIMO channel model is derived as a real matrix, which remains maximal multiplexing capacity and reduces the channel estimation complexity. Simulation results show that the proposed scheme has outstanding performance over Rician flat fading channel, and experimental system obtains four times the capacity through 4 antennas on both transmitter and receiver.


Game-Based Distributed Noncooperation Interference Coordination Scheme in Ultra-Dense Networks

张永昌

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0026

摘要 ( 434 ) PDF (1219 KB)( 87 )

参考文献 | 相关文章 | 计量指标

Ultra-dense networks (UDNs) is a promising solution to meet the exponential increase in mobile data traffic. But the ultra-dense deployment of cells inevitably brings complicated inter-cell interference ( ICI) and existing interference coordination scheme cannot be directly applied. To minimize the aggregate interference of each small cells, this paper formulates the problem as a distributed noncooperation game-based interference coordination scheme in ultra-dense networks considering the real demand rate of each small cell user equipment (SUE) and proves it to be a potential game. An improved no-regret learning algorithm is introduced to coverage to the Nash equilibrium (NE) of the formulated game. Simulation results show that the proposed scheme has better performance compared with existing schemes.


Low-complexity transmit antenna selection algorithm for massive MIMO

李新民, 李国民, 刘洋, 郭甜, 李璞

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0029

摘要 ( 433 ) PDF (397 KB)( 217 )

参考文献 | 相关文章 | 计量指标

Massive multiple input multiple output (MIMO) systems can increase capacity and reliability greatly. However, extremely high hardware costs and computational complexity lead to the demand for reasonable antenna selection. Aiming at the problem that the traditional antenna selection algorithm based on maximizing sum capacity has large complexity and worse bit error rate (BER) performance, a two-step selection algorithm is proposed, which selects a part of the antennas based on the norm-based antenna selection (NBS) firstly, and then selects the antenna based on maximizing capacity via convex optimization. The simulation results show that the improved algorithm has better BER performance than the traditional algorithms. At the same time, it reduces computational complexity greatly.


RFID Indoor positioning based on Semi-supervised Actor-Critic Co-training

李丽, 郑嘉利, 全艺璇, 林子涵, 李映超, 黄天星

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0030

摘要 ( 487 ) PDF (6579 KB)( 171 )

参考文献 | 相关文章 | 计量指标

For large-scale radio frequency identification ( RFID) indoor positioning system, the positioning scale is relatively large, with less labeled data and more unlabeled data, and it is easily affected by multipath and white noise. An RFID positioning algorithm based on semi-supervised actor-critic co-training (SACC) was proposed to solve this problem. In this research, the positioning is regarded as Markov decision-making process. Firstly, the actor-critic was combined with random actions and selects the unlabeled best received signal arrival intensity (RSSI) data by co-training of the semi-supervised. Secondly, the actor and the critic were updated by employing Kronecker-factored approximation calculate (K-FAC) natural gradient. Finally, the target position was obtained by co-locating with labeled RSSI data and the selected unlabeled RSSI data. The proposed method reduced the cost of indoor positioning significantly by decreasing the number of labeled data. Meanwhile, with the increase of the positioning targets, the actor could quickly select unlabeled RSSI data and updates the location model. Experiment shows that, compared with other RFID indoor positioning algorithms, such as twin delayed deep deterministic policy gradient (TD3), deep deterministic policy gradient (DDPG), and actor-critic using Kronecker-factored trust region ( ACKTR), the proposed method decreased the average positioning error respectively by 50.226%, 41.916%, and 25.004%. Meanwhile, the positioning stability was improved by 23.430%, 28.518%, and 38.631%.


Others

City-wide vehicle dispatching for multi-hop ridesharing package delivery

赵西雨, 张雪菲, 刘俊杰, 王怡宁

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0028

摘要 ( 643 ) PDF (1780 KB)( 110 )

参考文献 | 相关文章 | 计量指标




The city-wide ridesharing package delivery is becoming popular as it provides a convenience such as extra profits to the vehicle-s driver and high traffic efficiency to the city. The vehicle dispatching is a significant issue to improve the ridesharing efficiency in package delivery. The classic one-hop ridesharing package delivery requires the highly similar paths between the package and the vehicle given by the limited detour time, which depresses the ridesharing efficiency. To tackle this problem, a city-wide vehicle dispatching strategy for the multi-hop ridesharing package delivery was proposed, where a package is permitted to be delivered sequentially by different vehicles, until arriving the destination. The study formulates the vehicle dispatching as a maximum multi-dimensional bipartite matching problem with the goal of maximizing the total saving distance given by the limited detour time and ridesharing capacity. A multi-hop ridesharing vehicle dispatching algorithm was proposed to solve this problem by selecting the farthest reachable locker and multi-dimensional matching. Simulation results based on real vehicle dataset of Beijing demonstrate the effectiveness and efficiency of the proposed vehicle dispatching strategy.




Prediction of network attack profit path based on NAPG model

刘琨, 王辉, 申自浩

中国邮电高校学报(英文), 2020, 27 (5). doi： 10.19682/j.cnki.1005-8885.2020.0021

摘要 ( 492 ) PDF (1821 KB)( 105 )

参考文献 | 相关文章 | 计量指标

The network attack profit graph (NAPG) model and the attack profit path predication algorithm are presented herein to cover the shortage of considerations in attacker-s subjective factors based on existing network attack path prediction methods. Firstly, the attack profit is introduced, with the attack profit matrix designed and the attack profit matrix generation algorithm given accordingly. Secondly, a path profit feasibility analysis algorithm is proposed to analyze the network feasibility of realizing profit of attack path. Finally, an opportunity profit path and an optimal profit path are introduced with the selection algorithm and the prediction algorithm designed for accurate prediction of the path. According to the experimental test, the network attack profit path predication algorithm is applicable for accurate prediction of the opportunity profit path and the optimal profit path.