Self-Stylized Neural Painter

Conference
Qian Wang, Cai Guo, Hong-Ning Dai, Ping Li
SIGGRAPH Asia 2021 Posters, Tokyo, Japan, 2021
Publication year: 2021

This work introduces Self-Stylized Neural Painter (SSNP), a deep neural network that automatically creates stylized artworks in a stroke-by-stroke manner. Our SSNP consists of digit artist, canvas, style-stroke generator (SSG). By using SSG to generate style strokes, SSNP creates different styles paintings based on the given images. We design SSG as a three-player game based on a generative adversarial network to produce pure-color strokes that are crucial for mimicking the physical strokes by human artists. Furthermore, the digital artist adjusts the parameters of strokes (shape, size, transparency, and color) to reconstruct as many detailed contents of the reference image as possible to improve the fidelity.

MFF-AMD: Multivariate Feature Fusion for Android Malware Detection

Conference
Guangquan Xu, Meiqi Feng, Litao Jiao, Jian Liu, Hong-Ning Dai, Ding Wang, Emmanouil Panaousis, Xi Zheng
EAI CollaborateCom 2021
Publication year: 2021

Researchers have turned their focus on leveraging either dynamic or static features extracted from applications to train AI algorithms to identify malware precisely. However, the adversarial techniques have been continuously evolving and meanwhile, the code structure and application function have been designed in complex format. This makes Android malware detection more challenging than before. Most of the existing detection methods may not work well on recent malware samples. In this paper, we aim at enhancing the detection accuracy of Android malware through machine learning techniques via the design and development of our system called MFF-AMD. In our system, we first extract various features through static and dynamic analysis and obtain a multiscale comprehensive feature set. Then, to achieve high classification performance, we introduce the Relief algorithm to fuse the features, and design four weight distribution algorithms to fuse base classifiers. Finally, we set the threshold to guide MFF-AMD to perform static or hybrid analysis on the malware samples. Our experiments performed on more than 25,000 applications from the recent five-year dataset demonstrate that MFF-AMD can effectively detect malware with high accuracy.

Ground-to-UAV Communication Network: Stochastic Geometry-based Performance Analysis

Conference
Yalin Liu, Hong-Ning Dai, Muhammd Imran, Nidal Nasser
IEEE International Conference on Communications (ICC 2021)
Publication year: 2021

In this paper, we employ stochastic geometry to analyze ground-to-unmanned aerial vehicle (UAV) communications. We consider multiple UAVs to provide user-equipments (UEs) with uplink transmissions, where the distribution of UEs follows the Poisson Cluster process (PCP) and each UAV is dedicated to a specific cluster. In particular, we characterize the Laplace transform of the interference caused by multiple UEs in terms of the distribution of UEs as well as the transmission probability of each UE. We then derive analytical expressions of the successful transmission probability. We next conduct a comprehensive numerical analysis with consideration of different system parameters. The results show that four factors (i.e., the geographical surroundings, the transmission powers, the Signal-to-Interference-plus-Noise Ratio (SINR) thresholds, and the UAV height) have main influences on ground-to-UAV communications.

Ear in the Sky: Terrestrial Mobile Jamming to Prevent Aerial Eavesdropping

Conference
Qubeijian Wang, Yalin Liu, Hong-Ning Dai, Muhammad Imran, Nidal Nasser
The 2021 IEEE Global Communications Conference (GLOBECOM), 2021, Madrid, Spain (On-site & Virtual)
Publication year: 2021

The emerging unmanned aerial vehicles (UAVs) pose a potential security threat for terrestrial communications when UAVs can be maliciously employed as UAV-eavesdroppers to wiretap confidential communications. To address such an aerial security threat, we present a friendly jamming scheme named terrestrial mobile jamming (TMJ) to protect terrestrial confidential communications from UAV eavesdropping. In our TMJ scheme, a jammer moving along the protection area can emit jamming signals toward the UAV-eavesdropper so as to reduce the eavesdropping risk. We evaluate the performance of our scheme by analyzing a secrecy-capacity maximization problem subject to the legitimate connectivity and eavesdropping probability. In addition, we investigate the optimized position for the jammer as well as its jamming power. Simulation results verify the effectiveness of the proposed scheme.

Coverage Analysis of Blockchain-enabled Wireless IoMT Networks

Conference
Xuran Li, Hong-Ning Dai, Jie Tian, Dehuan Wan, Dengwang Li
2021 IEEE Globecom Workshops (GC Wkshps): Workshop on Scalable, Secure and Intelligent Blockchain for Future Networking and Communications
Publication year: 2021

The blockchain-enabled wireless Internet of medical things (IoMT) system has recently drawn extensive attention due to the provision of highly-secured healthcare services. However, multiple simultaneous transmissions in blockchain-enabled IoMT (BC-IoMT) networks can cause interference, consequently degrading the overall performance of the system. It is necessary to investigate the performance of BC-IoMT networks. In this paper, we propose a novel analytical framework for wireless BC-IoMT networks with consideration of both the spatial model on geographical random distribution of IoMT users and the temporal model on stochastic nature of data block transmission. With this framework, we derive a closed-form expression of the coverage probability in the wireless BC-IoMT network. Our extensive simulation results verify the accuracy of our theoretical analysis. From the analytical framework, we find that the path loss effect, the number of interfering medical sensor devices, and the threshold of successful transmission have a significant impact on the coverage performance of wireless BC-IoMT networks.

Connectivity Analysis of UAV-To-Satellite Communications in Non-Terrestrial Networks

Conference
Yalin Liu, Hong-Ning Dai, Ning Zhang
The 2021 IEEE Global Communications Conference (GLOBECOM), 2021, Madrid, Spain (On-site & Virtual)
Publication year: 2021

Non-terrestrial Networks (NTNs) refer to the networks, where either satellites or unmanned aerial vehicles (UAVs) are deployed to extend the current terrestrial networks for serving the growing mobile broadband and machine-type communications. With the advantages of UAVs’ flexibility and satellites’ global coverage, the solution of UAV-To-satellite communications (U2SC) can provide promising global communication services for the emerging NTNs. Previous literature has explored many potential directions of U2SC, including channel tracking, deployment design, and link analysis. However, as a vital role in system performance, the connectivity of U2SC has not been well investigated yet. This research gap motivates us to present an analytical model to evaluate the connectivity of U2SC. In particular, we first present the system model of the U2SC by considering the distribution model of UAVs, antenna models, and the path loss model. We then utilize stochastic geometry to derive a theoretical formulation of the successful connection probability of U2SC. The comprehensive numerical results are given to evaluate the received power, the interference, and the successful connection probability of U2SC and analyze the impacts of system parameters, such as the number of frequency carriers, the type of frequency bands, the number of UAVs, and the satellite altitude.

Complex Network Analysis of the Bitcoin Blockchain Network

Conference
Bishenghui Tao, Ivan Wang Hei Ho, Hong-Ning Dai
IEEE International Symposium on Circuits & Systems (ISCAS), Daegu, Korea, 2021 (On-site \& Virtual)
Publication year: 2021

On UAV-assisted Data Acquisition for Underwater IoT in Aquaculture Surveillance

Conference
Qubeijian Wang, Hong-Ning Dai, Qiu Wang and Mahendra K. Shukla
Proceedings of 5th International Conference on Maritime Technology and Engineering (MARTECH 2020)
Publication year: 2020

Underwater exploration activities have grown significantly due to the proliferation of underwater Internet of Things (UIoT). Recently, UIoT can potentially be used in fishing and aquaculture for environment surveillance such as water quality and temperature monitoring. UIoT nodes are distributed in the aquaculture region (i.e., under the water) to collect sensor data (i.e., surveillance data). However, to transmit sensor data from UIoT to remote onshore data processing center requires huge cost of deploying and maintaining communication infrastructures. In this paper, we propose an Unmanned Aerial Vehicles (UAVs)-assisted underwater data acquisition scheme by placing multiple sink nodes on the water surface to serve as intermediate relays between underwater sensors (IoT nodes) and UAVs. In our scheme, the sensor data is first transmitted via an acoustic-signal link to a buoyant sink node, which then forwards the data to a UAV via an electromagnetic link. In particular, we adopt a random placement of sink nodes. Since the path connectivity from an underwater sensor node to the UAV is crucial to guarantee reliable data acquisition tasks, we establish a theoretical framework to analyze the path connectivity via the intermediate sink node. Extensive simulation results validate the accuracy of the proposed analytical model. Moreover, our results also reveal the relationship between the path connectivity and other factors, such as sink node placements, antenna beamwidth of UAVs and wind speed.

LSH-based Collaborative Recommendation Method with Privacy-Preservation

Conference
Jiangmin Xu, Xuansong Li, Hao Wang, Hong-Ning Dai, Shunmei Meng
IEEE International Conference on Cloud Computing (IEEE CLOUD), 2020
Publication year: 2020

With the rapid development of cloud computing technology, massive services, and online information cause information overload. Collaborative Filtering (CF) is one of the most successful and widely used technologies in a personalized recommendation system to deal with information overload. However, traditional CF recommendation algorithms go through high time cost and poor real-time performance when dealing with large-scale behavior data. Moreover, most collaborative recommendation methods mainly focus on improving recommendation accuracy, while ignoring privacy preservation. In addition, the recommendation results of traditional CF recommendation algorithms are often too single, which could not meet the user’s diverse requirements. To solve these problems, this paper proposes a privacy-aware collaborative recommendation algorithm based on local sensitive hash (LSH) and factorization techniques. First, LSH is adopted to determine the nearest neighbor set of the target users, where a neighbor matrix for the target user can be generated. The matrix factorization technique is applied in the neighbor matrix to predict the missing ratings. Then the nearest neighbors can be determined based on the predicted ratings. Finally, predictions for the target user are made based on the neighborhood-based CF recommendation model and diversified recommendations are made for the target user. Experimental results show that the proposed algorithm can effectively improve the efficiency of recommendation on the premise of protecting the privacy of users.

Fused 3-Stage Image Segmentation for Pleural Effusion Cell Clusters

Conference
Sike Ma, Meng Zhao, Hao Wang, Fan Shi, Xuguo Sun, Shengyong Chen, Hong-Ning Dai
Proceedings of the 25th International Conference on Pattern Recognition (ICPR), 2020
Publication year: 2020

The appearance of tumor cell clusters in pleural effusion is usually a vital sign of cancer metastasis. Segmentation, as an indispensable basis, is of crucial importance for diagnosing, chemical treatment, and prognosis in patients. However, accurate segmentation of unstained cell clusters containing more detailed features than the fluorescent staining images remains to be a challenging problem due to the complex background and the unclear boundary. Therefore, in this paper, we propose a fused 3-stage image segmentation algorithm, namely Coarse segmentation-Mapping-Fine segmentation (CMF) to achieve unstained cell clusters from whole slide images. Firstly, we establish a tumor cell cluster dataset consisting of 107 sets of images, with each set containing one unstained image, one stained image, and one ground-truth image. Then, according to the features of the unstained and stained cell clusters, we propose a three-stage segmentation method: 1) Coarse segmentation on stained images to extract suspicious cell regions-Region of Interest (ROI); 2) Mapping this ROI to the corresponding unstained image to get the ROI of the unstained image (UI-ROI); 3) Fine Segmentation using improved automatic fuzzy clustering framework (AFCF) on the UI-ROI to get precise cell cluster boundaries. Experimental results on 107 sets of images demonstrate that the proposed algorithm can achieve better performance on unstained cell clusters with an F1 score of 90.40%.

Distance-Guided Mask Propagation Model for Efficient Video Object Segmentation

Conference
Jiajia Liu, Hong-Ning Dai, Bo Li, Gaozhong Tang
Proceedings of the 2020 International Joint Conference on Neural Networks (IJCNN 2020)
Publication year: 2020

Video object segmentation (VOS) is a significant yet challenging task in computer vision. In VOS, two challenging problems, including occlusions and distractions, are needed to be handled especially in multi-object videos. However, most existing methods have difficulty in efficiently tackling these two factors. To this end, a new semi-supervised VOS model, called Distance-Guided Mask Propagation Model (DGMPM), is proposed in this paper. Specifically, a novel embedding distance module, which is utilized to generate a soft cue for handling occlusions, is implemented by calculating distance difference between target features and the centers of foreground/background features. This non-parametric module that is based on global contrast between the target and reference features to detect target object regions even if occlusions still exist, is less sensitive to the feature scale. The prior knowledge of the previous frame is applied as spatial guidance in the decoder to reduce the effect of distractions. In addition, spatial attention blocks are designed to strengthen the network to focus on the target object and rectify the prediction results. Extensive experiments demonstrate that the proposed DGMPM achieves competitive performance on accuracy and runtime in comparison with state-of-the-art methods.

Cryptocurrencies Price Prediction Using- Weighted Memory Multi-Channels

Conference
Zhuorui Zhang, Junhao Zhou, Yanan Song, and Hong-Ning Dai
The 2020 International Conference on Blockchain and Trustworthy Systems, 2020
Publication year: 2020

After the invention of Bitcoin and a peer to peer electronic cash system based on the blockchain, the market of cryptocurrencies increases rapidly and attracts substantial interest from investors and researchers. Cryptocurrencies price volatility prediction is a challenging task owing to the high stochasticity of the markets. Econometric, machine learning and deep learning models are investigated to tackle the stochastic financial prices fluctuation and to improve the prediction accuracy. Although the introduction of exogenous factors such as macro-financial indicators and blockchain information helps the model prediction more accurately, the noise and effects from markets and political conditions are difficult to interpret and modelling. Inspired by the evidence of strong correlations among cryptocurrencies examined in previous studies, we originally propose a Weighted Memory Channels Regression (WMCR) model to predict the daily close price of cryptocurrencies. The proposed model receives time series of several heavyweight cryptocurrencies price and learns the interdependencies of them by recalibrating the weights of each sequence wisely. Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) components are exploited to establish memory and extract spatial and temporal features. Moreover, regularization methods including kernel regularizers and bias regularizers and Dropout method are exploited to improve the generalization ability of the proposed model. A battery of experiments are conducted in this paper. The results present that the WMCR model achieves the state-of-art performance and outperforms other baseline models.

Communication-Efficient Federated Learning in UAV-enabled IoV: A Joint Auction-Coalition Approach

Conference
Jer Shyuan Ng, Wei Yang Bryan Lim, Hong-Ning Dai, Zehui Xiong,Jianqiang Huang, Dusit Niyato, Xian-Sheng Hua, Cyril Leung, and Chunyan Miao
2020 IEEE Global Communications Conference
Publication year: 2020

Due to the advanced capabilities of the Internet of Vehicles (IoV) components such as vehicles, Roadside Units (RSUs) and smart devices as well as the increasing amount of data generated, Federated Learning (FL) becomes a promising tool given that it enables privacy-preserving machine learning. However, the performance of the FL suffers from the failure of communication links and missing nodes. Therefore, we propose the use of Unmanned Aerial Vehicles (UAVs) as wireless relays to facilitate the communications between the IoV components and the FL server and thus improving the accuracy of the FL. However, a single UAV may not have sufficient resources for all iterations of the FL process. In this paper, we present a joint auction-coalition formation framework. The joint auction-coalition formation algorithm is proposed to achieve a stable partition of UAV coalitions in which an auction scheme is applied. The auction scheme is designed to take into account the preferences of IoV components over heterogeneous UAVs. The simulation results show that the grand coalition, where all UAVs join a single coalition, is not always stable due to the profit-maximizing behavior of the UAVs. In addition, we show that as the cooperation cost of the UAVs increases, the UAVs prefer not to form any coalition.

Boosting UI Rendering in Android Applications

Conference
Subrota Kumar Mondal,Yu Pei, Hong-Ning Dai, H M Dipu Kabir, Jyoti Prakash Sahoo
Proceedings of 2020 IEEE 20th International Conference on Software Quality, Reliability and Security Companion (QRS-C)
Publication year: 2020

The Android operating system captures over 86% mobile OS market share and a large number of software developers are keen on developing applications for the Android platform. Many Android applications, however, suffer from the problem of slow UI rendering, thereby losing their competitive edge. To be able to address this problem, the developers first need to understand the underlying reasons. In this paper, we present an empirical study on reasons for slow UI rendering on the Android platform, with its focus on the impact of (poor) layout implementation on UI rendering. We also propose a taxonomy of existing techniques that might help tackle the problem and strategies for efficient layout implementation. Results from applying the strategies to sample applications demonstrate that they can help enhance the efficiency of UI rendering.

UAV-enabled Data Acquisition Scheme with Directional Wireless Energy Transfer

Conference
Yalin Liu, Hong-Ning Dai, Yuyang Peng, Hao Wang
International Conference on Embedded Wireless Systems and Networks (EWSN) (Poster session)
Publication year: 2019

In this paper, we exploit an Unmanned Aerial Vehicle (UAV) as a data collector which first transfers wireless energy to an Internet of Things (IoT) node who then sends back the data packets to UAV. In particular, we present a resource allocation scheme for the data acquisition task by minimizing the overall energy consumption. We further investigate two optional allocations for wireless energy transfer time and data transmitting power as well as the applicable conditions. Numerical results show the adaptability of our allocation scheme with the varied value of channel-fading parameter and data size level.

Portable Convolution Neural Networks for Traffic Sign Recognition in Intelligent Transportation Systems

Conference
Junhao Zhou, Hong-Ning Dai, Hao Wang
IEEE International Conference on Cyber Physical and Social Computing (CPSCom), 2019
Publication year: 2019

Abstract

Deep convolutional neural networks (CNN) have the strength in traffic-sign classification in terms of high accuracy. However, CNN models usually contains multiple layers with a large number of parameters consequently leading to a large model size. The bulky model size of CNN models prevents them from the wide deployment in mobile and portable devices in Intelligent Transportation Systems. In this paper, we design and develop a portable convolutional neural network (namely portable CNN) structure used for traffic-sign classification. This portable CNN model contains a stacked convolutional structure consisting of factorization and compression modules. We conducted extensive experiments to evaluate the performance of the proposed Portable CNN model. Experimental results show that our model has the advantages of smaller model size while maintaining high classification accuracy, compared with conventional CNN models.

Keywords

  • Convolutional neural networks
  • Portable
  • Model Compression
  • Intelligent Transportation Systems

Bibtex

@INPROCEEDINGS{JZhou:CPSCOM19, 
	author={Junhao Zhou and Hong-Ning Dai and Hao Wang}, 
	booktitle={IEEE International Conference on Cyber Physical and Social Computing (CPSCom)}, 
	title={Portable Convolution Neural Networks for Traffic Sign Recognition in Intelligent Transportation Systems}, 
	year={2019}, 
	volume={}, 
	number={}, 
	pages={1-6}, 
}

MHDT: A Deep-Learning-based Text Detection Algorithm for Unstructured Data in Banking

Conference
Shenglan Ma, Lingling Yang, Hao Wang, Hong Xiao, Hong-Ning Dai, Shuhan Cheng, Tongsen Wan
11th International Conference on Machine Learning and Computing, 2019
Publication year: 2019

Gold Price Forecast based on LSTM-CNN Model

Conference
Zhanhong He, Junhao Zhou, Hong-Ning Dai, Hao Wang
The 5th International Conference on Cloud and Big Data Computing (CBDCom), 2019
Publication year: 2019

Abstract

An accurate prediction is certainly significant in financial data analysis. Investors have used a series of econometric techniques on pricing, stock selection and risk management but few of them have found great success due to the fact that most of them only are purely based on a single scheme. Recent advances in deep learning methods have also demonstrated the outstanding performance in the fields of image recognition and sentiment analysis. In this paper, we originally propose a novel gold price forecast method based on the integration of Long Short-Term Memory Neural Networks (LSTM) and Convolutional Neural Networks (CNN) with Attention Mechanism (denoted to LSTM-Attention-CNN model). Particularly, the LSTM-Attention-CNN model consists of three components: the LSTM component, Attention Mechanism and the CNN component. The LSTM component enables to harness the sequential order of daily gold price. Meanwhile, the attention mechanism assigns different attention weights on the new encoding method from LSTM component to enhance the extraction of the temporal features. In addition, the CNN component enables to capture the local patterns and abstract the spatial features. Extensive experiments on real dataset collected from World Gold Council show that our proposed approach outperforms other conventional financial forecast methods.

Keywords

  • Long short-term memory
  • Convolutional neural network
  • Gold price prediction
  • Deep learning

Bibtex

@INPROCEEDINGS{ZHe:CBDCOM19, 
	author={Zhanhong He and Junhao Zhou and Hong-Ning Dai and Hao Wang}, 
	booktitle={The 5th International Conference on Cloud and Big Data Computing (CBDCom)}, 
	title={Gold Price Forecast based on LSTM-CNN Model}, 
	year={2019}, 
	volume={}, 
	number={}, 
	pages={1-6}, 
}

When Friendly Jamming Meets Wireless Energy Transfer

Conference
Qi Sun, Hong-Ning Dai, Qiu Wang, Xuran Li, Hao Wang
IEEE International Conference on Green Computing and Communications 2018, Halifax, Canada
Publication year: 2018

Friendly-jamming schemes can effectively reduce the eavesdropping risk in wireless networks by generating sufficient interference to prevent eavesdroppers from snooping confidential communications. However, this type of anti-eavesdropping schemes can also affect the normal communications due to the interference to legitimate users. On the other hand, Wireless Energy Transfer (WET) technology has received much attention recently since WET allows a node to obtain the energy from electromagnetic radiation. In this paper, we integrate the friendly-jamming scheme with WET. We call this scheme as Wireless-Jamming-Energy-Transfer (WJET). This scheme can translate the harmful interference radiated from jammers into the energy harvested by legitimate transmitters. In order to evaluate the effectiveness of this scheme, we establish an analytical model to analyze the transmission probability and the eavesdropping probability. Simulations verify that WJET scheme can simultaneously decrease the eavesdropping probability of eavesdroppers and increase the transmission probability of legitimate users. In addition, we investigate the density of jammers to achieve the optimal transmission probability according to various channel conditions, the density of transmitters and the transmission power of jammers.

Sentiment Analysis of Chinese Microblog Based on Stacked Bidirectional LSTM

Conference
Yue Lu, Junhao Zhou, Hong-Ning Dai, Hao Wang, Hong Xiao
The 15th International Symposium on Pervasive Systems, Algorithms and Networks (I-SPAN 2018)
Publication year: 2018

Recently, sentiment analysis on microblogs has received extensive attention recently. Most of previous studies focus on identifying sentiment orientation by encoding as many word properties as possible. However, most of them ignore the long-range dependencies of words (i.e., contextual features), which are essentially important in the sentiment analysis. In this paper, we propose a sentiment analysis method by incorporating Continuous Bag-of-Words (CBOW) model and Stacked Bidirectional long short-term memory (Stacked Bi-LSTM) model to enhance the performance of sentiment prediction. Firstly, a word embedding model, CBOW model, is employed to capture semantic features of words and transfer words into high dimensional word vectors. Secondly, we introduce Stacked Bi-LSTM model to conduct the feature extraction of sequential word vectors at a deep level. Finally, a binary softmax classifier utilizes semantic and contextual features to predict the sentiment orientation. Extensive experiments on real dataset collected from Weibo (i.e., one of the most popular Chinese microblogs) show that our proposed approach achieves better performance than other machine learning models.