Digital Library

Vol. 17, No. 3, Jun. 2021  Cover

1. A Federated Multi-Task Learning Model Based onAdaptive Distributed Data Latent Correlation Analysis
   Shengbin Wu, Yibai Wang
   Vol. 17, No. 3, pp. 441-452, Jun. 2021  
   https://doi.org/10.3745/JIPS.01.0075
   Details Full Text PubReader
   Keywords: Data distribution, Federated Multi-Task Learning, Rank Constraint, Underlying Structure
   Show / Hide Abstract

   Federated learning provides an efficient integrated model for distributed data, allowing the local training of different data. Meanwhile, the goal of multi-task learning is to simultaneously establish models for multiple related tasks, and to obtain the underlying main structure. However, traditional federated multi-task learning models not only have strict requirements for the data distribution, but also demand large amounts of calculation and have slow convergence, which hindered their promotion in many fields. In our work, we apply the rank constraint on weight vectors of the multi-task learning model to adaptively adjust the task’s similarity learning, according to the distribution of federal node data. The proposed model has a general framework for solving optimal solutions, which can be used to deal with various data types. Experiments show that our model has achieved the best results in different dataset. Notably, our model can still obtain stable results in datasets with large distribution differences. In addition, compared with traditional federated multi-task learning models, our algorithm is able to converge on a local optimal solution within limited training iterations.


2. Attentive Transfer Learning via Self-supervisedLearning for Cervical Dysplasia Diagnosis
   Jinyeong Chae, Roger Zimmermann, Dongho Kim, Jihie Kim
   Vol. 17, No. 3, pp. 453-461, Jun. 2021  
   https://doi.org/10.3745/JIPS.04.0214
   Details Full Text PubReader
   Keywords: Attention Learning, Cervical Dysplasia, Patch self-supervised Learning, Transfer Learning
   Show / Hide Abstract

   Many deep learning approaches have been studied for image classification in computer vision. However, there are not enough data to generate accurate models in medical fields, and many datasets are not annotated. This study presents a new method that can use both unlabeled and labeled data. The proposed method is applied to classify cervix images into normal versus cancerous, and we demonstrate the results. First, we use a patch selfsupervised learning for training the global context of the image using an unlabeled image dataset. Second, we generate a classifier model by using the transferred knowledge from self-supervised learning. We also apply attention learning to capture the local features of the image. The combined method provides better performance than state-of-the-art approaches in accuracy and sensitivity."


3. Effective Pre-rating Method Based on Users'Dichotomous Preferences and Average RatingsFusion for Recommender Systems
   Shulin Cheng, Wanyan Wang, Shan Yang, Xiufang Cheng
   Vol. 17, No. 3, pp. 462-472, Jun. 2021  
   https://doi.org/10.3745/JIPS.01.0076
   Details Full Text PubReader
   Keywords: Collaborative Filtering, data sparsity, Fusion Filling, Preference Matrix, Recommender System
   Show / Hide Abstract

   With an increase in the scale of recommender systems, users’ rating data tend to be extremely sparse. Some methods have been utilized to alleviate this problem; nevertheless, it has not been satisfactorily solved yet. Therefore, we propose an effective pre-rating method based on users’ dichotomous preferences and average ratings fusion. First, based on a user–item ratings matrix, a new user-item preference matrix was constructed to analyze and model user preferences. The items were then divided into two categories based on a parameterized dynamic threshold. The missing ratings for items that the user was not interested in were directly filled with the lowest user rating; otherwise, fusion ratings were utilized to fill the missing ratings. Further, an optimized parameter λ was introduced to adjust their weights. Finally, we verified our method on a standard dataset. The experimental results show that our method can effectively reduce the prediction error and improve the recommendation quality. As for its application, our method is effective, but not complicated.


4. Analysis of Impact on ERP CustomizationModule Using CSR Data
   Byung-Keun Yoo, Seung-Hee Kim
   Vol. 17, No. 3, pp. 473-488, Jun. 2021  
   https://doi.org/10.3745/JIPS.04.0215
   Details Full Text PubReader
   Keywords: Customization, ERP, Information System, Package Software, Project Cost Calculation
   Show / Hide Abstract

   "The enterprise resource planning (ERP) system is a standardized and advanced business process that many companies are implementing now-a-days through customization. However, it affects the efficiency of operations as these customizations are based on uniqueness. In this study, we analyzed the impact of customized modules and processing time on customer service request (CSR), by utilizing the stacked CSR data during the construction and operation of ERP, focusing on small and medium-sized enterprises (SMEs). As a result, a positive correlation was found between unit companies and the length of ERP implementation; ERP modules and the length of ERP implementation; ERP modules and unit companies; and the type of ERP implementation and ERP module. In terms of CSR, a comparison of CSR processing time of CBO (customized business object) module and STD (standard) module revealed that while the five modules did not display statistically significant differences, one module demonstrated a statistically very significant difference. In sum, the analysis indicates that the CBO-type CSR and its processing cost are higher than those of STD-type CSR. These results indicate that companies planning to implement an ERP system should consider the ERP module and their customization ratio and level. It not only gives the theoretical validity that should be considered as an indicator for decision making when ERP is constructed, but also its implications on the impact of processing time suggesting that the maintenance costs and project scheduling of ERP software must also be considered. This study is the first to present the degree of impact on the operation and maintenance of customized modules based on actual data and can provide a theoretical basis for applying SW change ratio in the cost estimation of ERP system maintenance."


5. An Offloading Scheduling Strategy with MinimizedPower Overhead for Internet of Vehicles Based onMobile Edge Computing
   Bo He, Tianzhang Li
   Vol. 17, No. 3, pp. 489-504, Jun. 2021  
   https://doi.org/10.3745/JIPS.01.0077
   Details Full Text PubReader
   Keywords: internet of vehicles, Minimizing Power Overhead, Mobile Edge Computing, Optimization Model, simulated annealing algorithm, Task offloading
   Show / Hide Abstract

   By distributing computing tasks among devices at the edge of networks, edge computing uses virtualization, distributed computing and parallel computing technologies to enable users dynamically obtain computing power, storage space and other services as needed. Applying edge computing architectures to Internet of Vehicles can effectively alleviate the contradiction among the large amount of computing, low delayed vehicle applications, and the limited and uneven resource distribution of vehicles. In this paper, a predictive offloading strategy based on the MEC load state is proposed, which not only considers reducing the delay of calculation results by the RSU multi-hop backhaul, but also reduces the queuing time of tasks at MEC servers. Firstly, the delay factor and the energy consumption factor are introduced according to the characteristics of tasks, and the cost of local execution and offloading to MEC servers for execution are defined. Then, from the perspective of vehicles, the delay preference factor and the energy consumption preference factor are introduced to define the cost of executing a computing task for another computing task. Furthermore, a mathematical optimization model for minimizing the power overhead is constructed with the constraints of time delay and power consumption. Additionally, the simulated annealing algorithm is utilized to solve the optimization model. The simulation results show that this strategy can effectively reduce the system power consumption by shortening the task execution delay. Finally, we can choose whether to offload computing tasks to MEC server for execution according to the size of two costs. This strategy not only meets the requirements of time delay and energy consumption, but also ensures the lowest cost.


6. Implementation of an Autostereoscopic Virtual 3DButton in Non-contact Manner Using SimpleDeep Learning Network
   Sang-Hee You, Min Hwang, Ki-Hoon Kim, Chang-Suk Cho
   Vol. 17, No. 3, pp. 505-517, Jun. 2021  
   https://doi.org/10.3745/JIPS.02.0155
   Details Full Text PubReader
   Keywords: AI, Deep Learning, Non-contact, Stereoscopic, Virtual Button, 3D
   Show / Hide Abstract

   This research presented an implementation of autostereoscopic virtual three-dimensional (3D) button device as non-contact style. The proposed device has several characteristics about visible feature, non-contact use and artificial intelligence (AI) engine. The device was designed to be contactless to prevent virus contamination and consists of 3D buttons in a virtual stereoscopic view. To specify the button pressed virtually by fingertip pointing, a simple deep learning network having two stages without convolution filters was designed. As confirmed in the experiment, if the input data composition is clearly designed, the deep learning network does not need to be configured so complexly. As the results of testing and evaluation by the certification institute, the proposed button device shows high reliability and stability.


7. Unidirectional Flow: A Survey on Networks,Applications, and Characteristic Attributes
   Laxmisha Rai
   Vol. 17, No. 3, pp. 518-536, Jun. 2021  
   https://doi.org/10.3745/JIPS.03.0159
   Details Full Text PubReader
   Keywords: Acyclic, Algorithms, Attributes, Flow, routing, Unidirectional
   Show / Hide Abstract

   Studies and applications related to unidirectional flow are gaining attention from researchers across disciplines in the recent years. Flow can be viewed as a concept, where the material, fluid, people, air, and electricity are moving from one node to another over a transportation network, water network, or through electricity distribution systems. Unlike other networks such as computer networks, most of the flow networks are visible and have strong material existence and are responsible for the flow of materials with definite shape and volume. The flow of electricity is also unidirectional, and also share similar features as of flow of materials such as liquids and air. Generally, in a flow network, every node in the network participates and contributes to the efficiency of the network. In this survey paper, we would like to evaluate and analyze the depth and application of the acyclic nature of unidirectional flow in several domains such as industry, biology, medicine, and electricity. This survey also provides, how the unidirectional flow and flow networks play an important role in multiple disciplines. The study includes all the major developments in the past years describing the key attributes of unidirectional flow networks, including their applications, scope, and routing methods.


8. A Survey of Automatic Code Generation fromNatural Language
   Jiho Shin, Jaechang Nam
   Vol. 17, No. 3, pp. 537-555, Jun. 2021  
   https://doi.org/10.3745/JIPS.04.0216
   Details Full Text PubReader
   Keywords: Naturalistic Programming, software engineering, Survey, Source Code Generation
   Show / Hide Abstract

   Many researchers have carried out studies related to programming languages since the beginning of computer science. Besides programming with traditional programming languages (i.e., procedural, object-oriented, functional programming language, etc.), a new paradigm of programming is being carried out. It is programming with natural language. By programming with natural language, we expect that it will free our expressiveness in contrast to programming languages which have strong constraints in syntax. This paper surveys the approaches that generate source code automatically from a natural language description. We also categorize the approaches by their forms of input and output. Finally, we analyze the current trend of approaches and suggest the future direction of this research domain to improve automatic code generation with natural language. From the analysis, we state that researchers should work on customizing language models in the domain of source code and explore better representations of source code such as embedding techniques and pre-trained models which have been proved to work well on natural language processing tasks.


9. A Video Expression Recognition Method Based onMulti-mode Convolution Neural Network andMultiplicative Feature Fusion
   Qun Ren
   Vol. 17, No. 3, pp. 556-570, Jun. 2021  
   https://doi.org/10.3745/JIPS.02.0156
   Details Full Text PubReader
   Keywords: facial expression recognition, Multi-Mode Deep Learning, Multiplicative Fusion, Optical Flow Method, Spatial Convolutional Neural Network, Time Convolutional Neural Network
   Show / Hide Abstract

   The existing video expression recognition methods mainly focus on the spatial feature extraction of video expression images, but tend to ignore the dynamic features of video sequences. To solve this problem, a multimode convolution neural network method is proposed to effectively improve the performance of facial expression recognition in video. Firstly, OpenFace 2.0 is used to detect face images in video, and two deep convolution neural networks are used to extract spatiotemporal expression features. Furthermore, spatial convolution neural network is used to extract the spatial information features of each static expression image, and the dynamic information feature is extracted from the optical flow information of multiple expression images based on temporal convolution neural network. Then, the spatiotemporal features learned by the two deep convolution neural networks are fused by multiplication. Finally, the fused features are input into support vector machine to realize the facial expression classification. Experimental results show that the recognition accuracy of the proposed method can reach 64.57% and 60.89%, respectively on RML and Baum-ls datasets. It is better than that of other contrast methods.


10. Secure Object Detection Based on Deep Learning
    Keonhyeong Kim, Im Young Jung
    Vol. 17, No. 3, pp. 571-585, Jun. 2021  
    https://doi.org/10.3745/JIPS.03.0161
    Details Full Text PubReader
    Keywords: Deep Learning, integrity, Object Detection, Privacy, robustness
    Show / Hide Abstract

    Applications for object detection are expanding as it is automated through artificial intelligence-based processing, such as deep learning, on a large volume of images and videos. High dependence on training data and a non-transparent way to find answers are the common characteristics of deep learning. Attacks on training data and training models have emerged, which are closely related to the nature of deep learning. Privacy, integrity, and robustness for the extracted information are important security issues because deep learning enables object recognition in images and videos. This paper summarizes the security issues that need to be addressed for future applications and analyzes the state-of-the-art security studies related to robustness, privacy, and integrity of object detection for images and videos.


11. Analyzing Customer Experience in Hotel ServicesUsing Topic Modeling
    Van-Ho Nguyen, Thanh Ho
    Vol. 17, No. 3, pp. 586-598, Jun. 2021  
    https://doi.org/10.3745/JIPS.04.0217
    Details Full Text PubReader
    Keywords: Customer Experience, Hotel Services, LDA, Text Mining, topic modeling
    Show / Hide Abstract

    Nowadays, users’ reviews and feedback on e-commerce sites stored in text create a huge source of information for analyzing customers’ experience with goods and services provided by a business. In other words, collecting and analyzing this information is necessary to better understand customer needs. In this study, we first collected a corpus with 99,322 customers’ comments and opinions in English. From this corpus we chose the best number of topics (K) using Perplexity and Coherence Score measurements as the input parameters for the model. Finally, we conducted an experiment using the latent Dirichlet allocation (LDA) topic model with K coefficients to explore the topic. The model results found hidden topics and keyword sets with high probability that are interesting to users. The application of empirical results from the model will support decision-making to help businesses improve products and services as well as business management and development in the field of hotel services.


12. Digital Signage System Based on IntelligentRecommendation Model in Edge Environment:The Case of Unmanned Store
    Kihoon Lee, Nammee Moon
    Vol. 17, No. 3, pp. 599-614, Jun. 2021  
    https://doi.org/10.3745/JIPS.02.0157
    Details Full Text PubReader
    Keywords: Correlation Analysis, Deep Learning, Digital Signage, Edge Computing, Recommended System
    Show / Hide Abstract

    This paper proposes a digital signage system based on an intelligent recommendation model. The proposed system consists of a server and an edge. The server manages the data, learns the advertisement recommendation model, and uses the trained advertisement recommendation model to determine the advertisements to be promoted in real time. The advertisement recommendation model provides predictions for various products and probabilities. The purchase index between the product and weather data was extracted and reflected using correlation analysis to improve the accuracy of predicting the probability of purchasing a product. First, the user information and product information are input to a deep neural network as a vector through an embedding process. With this information, the product candidate group generation model reduces the product candidates that can be purchased by a certain user. The advertisement recommendation model uses a wide and deep recommendation model to derive the recommendation list by predicting the probability of purchase for the selected products. Finally, the most suitable advertisements are selected using the predicted probability of purchase for all the users within the advertisement range. The proposed system does not communicate with the server. Therefore, it determines the advertisements using a model trained at the edge. It can also be applied to digital signage that requires immediate response from several users.


13. Strategy for Task Offloading of Multi-user andMulti-server Based on Cost Optimization inMobile Edge Computing Environment
    Yanfei He, Zhenhua Tang
    Vol. 17, No. 3, pp. 615-629, Jun. 2021  
    https://doi.org/10.3745/JIPS.01.0078
    Details Full Text PubReader
    Keywords: Cost Optimization, Distributed Computing, game theory, Mobile Edge Computing, Multi MEC Servers, Nash equilibrium, Task offloading
    Show / Hide Abstract

    With the development of mobile edge computing, how to utilize the computing power of edge computing to effectively and efficiently offload data and to compute offloading is of great research value. This paper studies the computation offloading problem of multi-user and multi-server in mobile edge computing. Firstly, in order to minimize system energy consumption, the problem is modeled by considering the joint optimization of the offloading strategy and the wireless and computing resource allocation in a multi-user and multi-server scenario. Additionally, this paper explores the computation offloading scheme to optimize the overall cost. As the centralized optimization method is an NP problem, the game method is used to achieve effective computation offloading in a distributed manner. The decision problem of distributed computation offloading between the mobile equipment is modeled as a multi-user computation offloading game. There is a Nash equilibrium in this game, and it can be achieved by a limited number of iterations. Then, we propose a distributed computation offloading algorithm, which first calculates offloading weights, and then distributedly iterates by the time slot to update the computation offloading decision. Finally, the algorithm is verified by simulation experiments. Simulation results show that our proposed algorithm can achieve the balance by a limited number of iterations. At the same time, the algorithm outperforms several other advanced computation offloading algorithms in terms of the number of users and overall overheads for beneficial decision-making.


14. Evaluation of Predictive Models for Early Identificationof Dropout Students
    JongHyuk Lee, Mihye Kim, Daehak Kim, Joon-Min Gil
    Vol. 17, No. 3, pp. 630-644, Jun. 2021  
    https://doi.org/10.3745/JIPS.04.0218
    Details Full Text PubReader
    Keywords: Educational Data Analysis, Student Dropout, Predictive model
    Show / Hide Abstract

    Educational data analysis is attracting increasing attention with the rise of the big data industry. The amounts and types of learning data available are increasing steadily, and the information technology required to analyze these data continues to develop. The early identification of potential dropout students is very important education is important in terms of social movement and social achievement. Here, we analyze educational data and generate predictive models for student dropout using logistic regression, a decision tree, a naïve Bayes method, and a multilayer perceptron. The multilayer perceptron model using independent variables selected via the variance analysis showed better performance than the other models. In addition, we experimentally found that not only grades but also extracurricular activities were important in terms of preventing student dropout.


15. An Automatic Urban Function District Division MethodBased on Big Data Analysis of POI
    Hao Guo, Haiqing Liu, Shengli Wang, Yu Zhang
    Vol. 17, No. 3, pp. 645-657, Jun. 2021  
    https://doi.org/10.3745/JIPS.04.0219
    Details Full Text PubReader
    Keywords: Big Data Analysis, Dichotomy Method, Mean Shift Algorithm, POI Data, Urban Function District
    Show / Hide Abstract

    Along with the rapid development of the economy, the urban scale has extended rapidly, leading to the formation of different types of urban function districts (UFDs), such as central business, residential and industrial districts. Recognizing the spatial distributions of these districts is of great significance to manage the evolving role of urban planning and further help in developing reliable urban planning programs. In this paper, we propose an automatic UFD division method based on big data analysis of point of interest (POI) data. Considering that the distribution of POI data is unbalanced in a geographic space, a dichotomy-based data retrieval method was used to improve the efficiency of the data crawling process. Further, a POI spatial feature analysis method based on the mean shift algorithm is proposed, where data points with similar attributive characteristics are clustered to form the function districts. The proposed method was thoroughly tested in an actual urban case scenario and the results show its superior performance. Further, the suitability of fit to practical situations reaches 88.4%, demonstrating a reasonable UFD division result.


16. Survey on Data Deduplication in Cloud StorageEnvironments
    Won-Bin Kim, Im-Yeong Lee
    Vol. 17, No. 3, pp. 658-673, Jun. 2021  
    https://doi.org/10.3745/JIPS.03.0160
    Details Full Text PubReader
    Keywords: Date Deduplication, Cloud Storage, Encryption, Security
    Show / Hide Abstract

    Data deduplication technology improves data storage efficiency while storing and managing large amounts of data. It reduces storage requirements by determining whether replicated data is being added to storage and omitting these uploads. Data deduplication technologies require data confidentiality and integrity when applied to cloud storage environments, and they require a variety of security measures, such as encryption. However, because the source data cannot be transformed, common encryption techniques generally cannot be applied at the same time as data deduplication. Various studies have been conducted to solve this problem. This white paper describes the basic environment for data deduplication technology. It also analyzes and compares multiple proposed technologies to address security threats.