International Journal of Computational Science and Engineering (99 papers in press)

Regular Issues

• A multi-objective optimisation multicast routing algorithm with diversity rate in cognitive wireless mesh networks  
  by Zhufang Kuang 
  Abstract: Cognitive Wireless Mesh Networks (CWMNs) were developed to improve the usage ratio of the licensed spectrum. Since the spectrum opportunities for users vary over time and location, enhancing the spectrum effectiveness is a goal and also a challenge for CWMNs. Multimedia applications have recently generated much interest in CWMNs supporting Quality-Of-Service (QoS) communications. Multicast routing and spectrum allocation is an important challenge in CWMNs. In this paper, we design an effective multicast routing algorithm based on diversity rate with respect to load balancing and the number of transmissions for CWMNs. A Load Balancing wireless links weight computing function and computing algorithm based on Diversity Rate (LBDR) are proposed, and a load balancing Channel and Rate Allocating algorithm based on Diversity Rate (CRADR) is proposed. On this basis, a Load balancing joint Multicast Routing, channel and Rate allocation algorithm based on Diversity rate with QoS constraints for CWMNs (LMR2D) is proposed. Balancing the load of node and channel, and minimising the number of transmissions of multicast tree are the objectives of LMR2D. Firstly, LMR2D computes the weight of wireless links using LBDR and the Dijkstra algorithm for constructing the load balancing multicast tree step by step. Secondly, LMR2D uses CRADR to allocate channel and rate of its to links, which is based on the Wireless Broadcast Advantage (WBA). Simulation results show that LMR2D can achieve the expected goal. Not only can it balance the load of node and channel, but also it needs fewer transmissions for multicast tree.
  Keywords: cognitive wireless mesh networks; multicast routing; spectrum allocation; load balanced; diversity rate.
  
  
• Graffiti-writing recognition with fine-grained information  
  by Jiashuang Xu, Jiashuang Zhangjie Fu 
  Abstract: Contactless HCI (Human-Computer Interaction) has become a new trend due to the springing up of the novel intelligent terminals. The existing interaction systems usually adopt depth cameras, motion controller, and radiofrequency devices. The common drawback of the above approaches is that all the participants are required to obey the unistroke writing standard for data acquisition. The uniformity of the writing rule simplifies the data acquisition stage, but it breaks the integrity of the handwriting system. In practice, the writing habits vary among people. It is observed that eight capitalised letters of the alphabet possess more than one writing pattern. Thus, we are motivated to propose a more adaptive, contactless graffiti-writing recognition system with CSI (Channel State Information) derived from Wi-Fi signals. The discrete wavelet transform is used for denoising. We choose a sliding window to calculate the MAD (Mean Absolute Deviation)to detect the start and end points. We extract the unique CSI waveform caused by writing action to represent each letter. To cater for more users writing customs and improve the universality of the system, we train separate HMMs (Hidden Markov Model) for the eight letters and conduct cross-validation for testing. The average detection accuracy reaches 94.5%. The average recognition accuracy for the 26-letter model is 85.96% when the number of the training sample is 100 from five subjects. The real-time recognition efficiency measured by characters per minute is 11.97(= 31/155.24 s).
  Keywords: air-write recognition; wireless sensing; channel state information.
  
  
• Experimental investigation and CFD simulation of power consumption for mixing in the gyro shaker  
  by P.A. Abdul Samad, P.R. Shalij 
  Abstract: The better mixing of ingredients is the key to improving the quality of the process in the manufacturing of several products. The gyro shaker is a dual rotation mixer commonly used for mixing highly viscous fluids. In this work, CFD simulation for the multiphase mixing in the gyro shaker is carried out for obtaining numerical solutions. Simulations of three different mixing models, namely Eulerian granular model, mixture model and volume of fluid (VOF) model are compared. Reynolds number and power number based on characteristic velocity were derived for the gyro shaker. Experiments were conducted to validate the mixing power by simulation using torque method and viscous dissipation method. The viscous dissipation method for mixing power demonstrates a smaller deviation from the experimental data than torque method. Among the three simulation models, the multiphase mixture model shows the minimum variation of the experimental data. A comparison of the flow fields of the different mixing models is also carried out.
  Keywords: computational fluid dynamics; characteristic velocity; Eulerian granular; gyro shaker; mixture model; multiphase; power consumption; power number; viscous dissipation; VOF; volume of fluid.
  
  
• Research on product design knowledge organisation model based on granularity principle  
  by Youyuan Wang, Weiwei Qian, Lu Zhao 
  Abstract: In order to solve the problem of weak discernibility relation between the demand of knowledge in the process of product design, a knowledge organisation model based on the granularity principle is put forward. The paper applies knowledge unit and knowledge point to describe product design knowledge, adopts the granularity principle to perform the granulation tissue of product design knowledge, monitors the classification, association and inference of knowledge points according to task requirements and structures, formalises the related knowledge, and ultimately provides knowledge service in the form of knowledge unit. Through the analysis of a case, the method is proven to be effective to improve the relevance of knowledge and to improve the efficiency of knowledge service.
  Keywords: knowledge organisation; granularity principle; product design.
  
  
• A novel clustering algorithm based on the deviation factor model  
  by Jungan Chen, Chen Yinyin, Yang Dongyong 
  Abstract: For classical clustering algorithms, it is difficult to find clusters that have non-spherical shapes or varied size and density. In view of this, many methods have been proposed in recent years to overcome this problem, such as introducing more representative points per cluster, considering both interconnectivity and closeness, and adopting the density-based method. However, the density defined in DBSCAN is decided by minPts and Eps, and it is not the best solution to describe the data distribution of one cluster. In this paper, a deviation factor model is proposed to describe the data distribution and a novel clustering algorithm based on artificial immune system is presented. The experimental results show that the proposed algorithm is more effective than DBSCAN, k-means, etc.
  Keywords: clustering algorithm; DBSCAN; artificial immune system.
  DOI: 10.1504/IJCSE.2019.10022775
   
  
• Multi-keywords carrier-free text steganography method based on Chinese Pinyin  
  by Yuling Liu, Jiao Wu, Guojiang Xin 
  Abstract: By combining big data with the characteristics of steganography, carrier-free steganography was proposed to resist all the steganalysis attacks. A novel method named multi-keywords carrier-free text steganography method, based on Chinese Pinyin, is introduced in this paper. In the proposed method, the hidden tags are selected from the Pinyin combinations of two words. In the process of information hiding, the POS (Part of Speech) is used for hiding the number of keywords. Also, the redundancy of hidden tags in extraction process is eliminated by ensuring the uniqueness of each hidden tag in each stego-text. Meanwhile, the way of joint retrieval is used for hiding multi-keywords. Experimental results show that the proposed method has good performance in the hiding capacity, the success rate of hiding, the extraction accuracy and the time efficiency with appropriate hidden tags and large scale of the big text data.
  Keywords: carrier-free steganography; big text data; multi-keywords; Chinese Pinyin; POS tagging.
  
  
• Collaborative filtering-based recommendation system for big data  
  by Jian Shen, Tianqi Zhou, Lina Chen 
  Abstract: The collaborative filtering algorithm is widely used in the recommendation system of e-commerce websites (Wong et al. 2016), which are based on the analysis of a large number of users' historical behaviour data, so as to explore the users' interest and recommend the appropriate products to users. In this paper, we focus on how to design a reliable and highly accurate algorithm for movie recommendation. It is worth noting that the algorithm is not limited to film recommendation, but can be applied in many other areas of e-commerce. In this paper, we use Java language to implement a movie recommendation system in Ubuntu system. Benefitting from the MapReduce framework and the recommendation algorithm based on items, the system can handle large data sets. The experimental results show that the system can achieve high efficiency and reliability in large datasets.
  Keywords: big data; collaborative filtering; e-commerce; movie recommendation; MapReduce framework.
  
  
• Communication optimisation for intermediate data of MapReduce computing model  
  by Yunpeng Cao, Haifeng Wang 
  Abstract: MapReduce is a typical computing model for processing and analysis of big data. MapReduce computing job produces a large amount of intermediate data after Map phase. Massive intermediate data results in a large amount of intermediate data communication across rack switches in the Shuffle process of MapReduce computing model, which degrades the performance of heterogeneous cluster computing. In order to optimise the intermediate data communication performance of Map-intensive jobs, the characteristics of pre-running scheduling information of MapReduce computing jobs are extracted, and job classification is realised by machine learning. The jobs of active intermediate data communication are mapped into a rack to keep the communication locality of intermediate data. The jobs with inactive communication are deployed to the nodes sorted by computing performance. The experimental results show that the proposed communication optimisation scheme has a good effect on Shuffle-intensive jobs, and can reach 4-5%. In the case of a larger amount of input data, the communication optimisation scheme is robust and can adapt to heterogeneous cluster. In the case of a multi-user application scene, the intermediate data communication can be reduced by 4.1%.
  Keywords: MapReduce computing model; big data processing; communication optimisation; intermediate data; machine learning.
  
  
• A routing strategy with energy optimisation based on community in mobile social networks  
  by Gaocai Wang, Nao Wang, Ying Peng, Shuqiang Huang 
  Abstract: In current mobile networks, usage has drastically shifted from mobile users base station end-to-end communication to message/content retrieval among mobile users, which forms a so-called mobile social network. Usually, in a mobile social network, the movement feature of the mobile users has social aggregation characteristics, and the same mobile user who visits different communities forms a connected network based on community. This paper studies the energy consumption optimisation problem of message delivery based on the social characteristics of mobile users. The paper proposes an optimal energy efficiency routing strategy based on community, which minimises the network energy consumption with a given delay. Firstly, the expected energy consumption and delay of message delivery in a connected network are obtained through Markov chain. Then a comprehensive cost function for message delivery from a source node to a destination node is designed, which is combined with energy consumption and delay. Thus we obtain the optimisation function for delivering a message of relay to comprehensive cost. Further, the reward function of relay is given. Finally, the optimal expected reward of optimal relay is achieved using the optimal stopping theory for realising the optimal energy efficiency routing strategy. In simulations, the average energy consumption, the average delay and the average de-livery ratio of the routing optimisation strategy in this paper are compared with those of other routing strategies in related literature. The results show that the strategy proposed by this paper has smaller average energy consumption, smaller average delay and bigger average delivery ratio, and better energy consumption optimisation results can be achieved.
  Keywords: mobile social networks; optimal energy efficiency routing; community; optimal stopping; optimal relay.
  
  
• Super-sampling by learning-based super-resolution  
  by Ping Du, Jinhuan Zhang, Jun Long 
  Abstract: In this paper, we present a novel problem of intelligent image processing, which is how to infer a finer image in terms of intensity levels for a given image. We explain the motivation for this effort and present a simple technique that makes it possible to apply the existing learning-based super-resolution methods to this new problem. As a result of the adoption of the intelligent methods, the proposed algorithm needs notably little human assistance. We also verify our algorithm experimentally in the paper.
  Keywords: texture synthesis; super-resolution; image manifold.
  DOI: 10.1504/IJCSE.2019.10020177
   
  
• An evolutionary algorithm for finding optimisation sequences: proposal and experiments  
  by João Fabrício Filho, Luis Gustavo Araujo Rodriguez, Anderson Faustino Da Silva 
  Abstract: Evolutionary algorithms are metaheuristics for solving combinatorial and optimisation problems. A combinatorial problem, important in the context of software development, consists in selecting code transformations that must be used by the compiler while generating the target code. The objective of this paper is to propose and evaluate an evolutionary algorithm that is capable of finding an efficient sequence of optimising transformations, which will be used while generating the target code. The results indicate that it is efficient to find good transformation sequences, and a good option to generate databases for machine learning systems.
  Keywords: evolutionary algorithms; code optimisation; iterative compilation; machine learning.
  
  
• Energy replenishment optimisation via density-based clustering  
  by Xin Gu, Jun Peng, Yijun Cheng, Xiaoyong Zhang, Kaiyang Liu 
  Abstract: This paper investigates a density-based clustering approach to achieve efficient energy replenishment in wireless rechargeable sensor networks (WRSNs). Sensor nodes with charging request are divided into several clusters. Some of them are selected as head nodes, adopting a mobile charger to visit. The rest are arranged to the closest head nodes. Then the mobile charger serves all nodes in the same cluster simultaneously. Different from other clustering algorithms, our proposed clustering approach selects the head nodes with high local density. The distance between high-density nodes is also taken into consideration, effectively reducing the charging delay. Simulation results show that our proposed clustering approach can achieve optimal cluster results. Moreover, compared with two other cluster-based charging methods, the charging delay and travel distance can be reduced by our proposed clustering approach, in both dense and sparse deployment scenarios.
  Keywords: wireless rechargeable sensor networks; clustering; mobile charger; wireless energy transfer; charging delay.
  
  
• Evolutionary ant colony algorithm using firefly-based transition for solving vehicle routing problems  
  by Rajeev Goel, Raman Maini 
  Abstract: In this paper, we propose an evolutionary optimisation algorithm that adapts the advantages of ant colony optimisation and firefly optimisation algorithms to solve the vehicle routing problem and its variants. Firefly optimisation (FA) based transition rule along with pheromone shaking rule is used to escape local optima. Whereas the multi-modal nature of FA helps in exploring the search space, pheromone shaking avoids the stagnation of pheromone deposit on the exploited paths. This is expected to improve the working of ant colony system (ACS). Performance of the proposed algorithm has been compared with the performance of some of other available meta-heuristic approaches currently being used for solving vehicle routing problems on some benchmark problems. Results show the consistency of the proposed approach. Moreover, its convergence rate is faster and the obtained solutions are closer to optimal compared with solutions obtained using certain other existing meta-heuristic approaches. The results also demonstrate the effectiveness of the presented algorithm over other existing FA-based algorithms for solving vehicle routing problems.
  Keywords: ant colony optimisation; evolutionary algorithms; firefly optimisation; vehicle routing problems.
  DOI: 10.1504/IJCSE.2020.10022039
   
  
• An integrated ambient intelligence system for a smart lab environment  
  by Dat Do, Scott King, Alaa Sheta, Thanh Pham 
  Abstract: The goals of the ambient intelligence system are not only to enhance the way people communicate with the surrounding environment but also to advance safety measures and enrich human lives. In this paper, we introduce an Integrated Ambient Intelligence System (IAmIS) to perceive the presence of people, identify them, determine their locations, and provide suitable interaction with them. The proposed framework can be applied in various application domains such as a smart house, authorisation, surveillance, crime prevention, and many others. The proposed system has five components: body detection and tracking, face recognition, controller, monitor system, and interaction modules. The system deploys RGB cameras and Kinect depth sensors to monitor human activity. The developed system is designed to be fast and reliable for indoor environments. The proposed IAmIS can interact directly with the environment or communicate with humans acting on the environment. Thus, the system behaves as an intelligent agent. The system has been deployed in our research lab and can recognise lab members and guests to the lab as well as track their movements and have interactions with them depending on their identity and location within the lab.
  Keywords: ambient intelligence system; awareness system; object tracking; face recognition; body tracking; Kinect.
  
  
• An internet-of-things based security scheme for healthcare environment for robust location privacy  
  by Aakanksha Tewari, Brij Gupta 
  Abstract: Recently, various applications of the internet of things have been developed for the healthcare sector. Our contribution is to provide a secure and low-cost environment for the IoT devices in healthcare. The main goal is to make patients lives easier and more comfortable by providing them with more effective treatments. Nevertheless, we also intend to address the issues related to location privacy and security due to the deployment of IoT devices. We have proposed a very simple mutual authentication protocol, which provides strong location privacy by using one-way hash, pseudo-random number generators and bitwise operations. Strong location privacy is a key factor while ensuring healthcare security. We can enforce this property by making sure that tags in the network are indistinguishable and the protocol ensures forward secrecy. The security strength of our protocol is verified through a formal proof model for location privacy.
  Keywords: internet of things; location privacy; RFID; mutual authentication; forward secrecy; indistinguishability.
  
  
• A fuzzy controller for an adaptive VNFs placement in 5G network architecture  
  by Sara Retal, Abdellah Idrissi 
  Abstract: In cloud computing, computation and memory resources are becoming a relevant growing business. On the other hand, mobile network architecture faces many hurdles, including lack of flexibility for providing enhanced services and distributed architecture, and expensive cost to provide a network topology that meets the users' equipment (UE) needs. To cope with these problems, cloud computing is used in mobile telecommunications market thanks to network functions virtualisation. In our paper, we develop a fuzzy controller to support virtual network functions placement and provide an adaptive solution to manage and organise the network. Our approach enables the solution to adapt to UE mobility and needs in terms of quality of experience. Furthermore, it minimises serving gateways relocation cost and the path between UEs and packet data network gateways, taking into account the resource capacities. The experimental results show that our approach provides good results compared with the literature methods.
  Keywords: cloud computing; virtual network functions placement; adaptive placement; fuzzy controller; multi-objective optimisation.
  DOI: 10.1504/IJCSE.2019.10018399
   
  
• The key user discovery model based on user importance calculation  
  by Lei Zhang, Dandan Jiang, Ruirong Xue, Yawen Yi, Xiangfeng Luo 
  Abstract: Recently, more and more users publish their views on events in social media. Identifying influential users in social media and calculating the importance of users can help to analyse the impact of hot events or enterprise products in the real world. A method based on attribute analysis selects relatively simple characteristics without digging into the event-targeted properties; the network-based analysis method only uses the user behaviour relation or the content association relation to build a network, which does not take the user attributes into consideration and cannot effectively calculate the user importance. This paper proposes a multi-angle user importance calculation method with event-specificity. The overall importance of a user is measured by taking into account the four levels within the user layer, the fan layer, the micro-blog layer, and the event layer. Experimental results show that our method can effectively calculate the importance of users.
  Keywords: key user discovery; multi-layer; social media.
  
  
• Event-triggered fault estimation for stochastic state-delay systems against adversaries with application to a DC motor system  
  by Yunji Li, Yi Gao, Quansheng Liu, Li Peng 
  Abstract: This paper is concerned with the problem of fault estimation for stochastic state-delay systems subject to adversaries under an event-triggered data-transmission framework. An adversarial fault estimator is designed for remote state and fault estimation simultaneously. Furthermore, a sufficient condition is provided for exponential stability in the mean-squared of the proposed event-triggered fault estimator. The corresponding event-triggered sensor data transmission scheme is to reduce the overall communication burden. Finally, an application example on a DC motor system is presented and the benefits of the obtained theoretical results are demonstrated by comparative experiments.
  Keywords: fault estimation; event-triggered data-transmission scheme; time delays.
  
  
• Dependence structure between bitcoin price and its influence factors  
  by Weili Chen, Zibin Zheng, Mingjie Ma, Jiajing Wu, Jiaquan Yao, Yuren Zhou 
  Abstract: Bitcoin is a decentralised digital currency that has attracted growing interest over recent years. Much research from different subjects emerged because bitcoin is a multidisciplinary product. Among all these studies, the interpretation of the drastic fluctuation of the bitcoin price attracts a great attention. Many influence factors of the bitcoin price were found. However, research seldom reveals the dependence structure between price and its influence factors. By selecting 10 interpretable influence factors from the bitcoin network and using copula theory, we find that the bitcoin price has different correlation structures with its influence factors. These findings provide new insights into the behaviour of miners, users, and coins in the bitcoin system, thus leading to meaningful implications for policymakers, investors and risk managers dealing with bitcoin and other cryptocurrencies.
  Keywords: bitcoin; price fluctuation; influence factor; dependency structure; copula function.
  DOI: 10.1504/IJCSE.2019.10018973
   
  
• Software design of monitoring and flight simulation for UAV swarms based on OSGEarth  
  by Meimin Wu, Yuxiang Xiao, Qian Bi 
  Abstract: Real-time monitoring of Unmanned Aerial Vehicle (UAV) swarms is critical for flight safety. In order to monitor the position and working condition of UAV intuitively, we propose a three-dimensional (3D) monitoring software for UAV swarms based on OpenSceneGraph Earth. The software is built on platform + plug-ins architecture. The flight scene is constructed via 3D visualisation. UAV nodes are updated and moved in the flight scene when data is received in real time. Meanwhile, in order to decrease the cost and improve the work efficiency in the development and performance verification of UAV swarms, the simulation platform for UAV swarms is designed. The swarm behaviour algorithm is pre-designed in a Python file, which will be read to parse the position data and display the flight scene. The software has been successfully applied to monitor the flight of UAV swarms. Excellent accuracy and reliability are demonstrated.
  Keywords: OSGEarth; UAV swarms; real-time monitoring; 3D visualisation; swarm simulation.
  DOI: 10.1504/IJCSE.2019.10024269
   
  
• Improved quantum secret sharing scheme based on GHZ states  
  by Ming-Ming Wang, Zhi-Guo Qu, Lin-Ming Gong 
  Abstract: With the rapid progress of quantum cryptography, secret sharing has been developed in the quantum setting for achieving a high level of security, which is known as quantum secret sharing (QSS). The first QSS scheme was proposed by Hillary et al. in 1999 [Phys. Rev. A 59, 1829 (1999)] based on entangled Greenberger-Horne-Zeilinger (GHZ) states. However, only 50% of the entangled quantum states are effective for eavesdropping detection and secret splitting in the original scheme. In this paper, we introduce a possible method, called measurement-delay strategy, to improve the qubit efficiency of the GHZ-based QSS scheme. By using this method, the qubit efficiency of the improved QSS scheme can reach 100% for both security detection and secret distribution. The improved QSS scheme can be implemented experimentally based on current technologies.
  Keywords: quantum secret sharing; efficiency; security; GHZ state.
  
  
• Analysing research collaboration through co-authorship networks in a big data environment: an efficient parallel approach  
  by Carlos Roberto Valêncio, José Carlos De Freitas, Rogéria Cristiane Gratão De Souza, Leandro Alves Neves, Geraldo Francisco Donegá Zafalon, Angelo Cesar Colombini, William Tenório 
  Abstract: Bibliometry is the quantitative study of scientific productions and enables the characterisation of scientific collaboration networks. However, with the development of science and the increase of scientific production, large collaborative networks are formed, which makes it difficult to extract bibliometrics. In this context, this work presents an efficient parallel optimisation of three bibliometrics for co-authorship network analysis using multithread programming: transitivity, average distance, and diameter. Our experiments found that the time wasted to calculate the transitivity value using the sequential approach grows 4.08 times faster than the parallel proposed approach when the size of co-authorship network grows. Similarly, the time wasted to calculate the average distance and diameter values using the sequential approach grows 5.27 times faster than the parallel proposed approach when the size of co-authorship network grows. In addition, we reported relevant values of speed up and efficiency for the developed algorithms.
  Keywords: bibliometrics; graphs; knowledge extraction; co-authorship network; NoSQL; parallel computing.
  DOI: 10.1504/IJCSE.2019.10020390
   
  
• Design of fault-tolerant majority voter for error-resilient TMR targeting micro- to nano-scale logic  
  by Mrinal Goswami, Subrata Chattopadhyay, Shiv Bhushan Tripathi, Bibhash Sen 
  Abstract: The shrinking size of transistors for satisfying the increasing demand for higher density and low power has made the VLSI circuits more vulnerable to faults. Therefore, new circuits in advanced VLSI technology have forced designers to use fault-tolerant techniques in safety-critical applications. Also, the presence of some faults (not permanent) due to the complexity of the nanocircuit or its interaction with software results in malfunctioning of circuits. The fault-tolerant scheme, where majority voter plays the core role in triple modular redundancy (TMR), is being implemented increasingly in digital systems. This work aims to implement a different fault-tolerant scheme of majority voter for the implementation of TMR using quantum-dot cellular automata (QCA), which is a viable alternative nanotechnology to CMOS VLSI. The fault-masking ability of various voter designs has been analysed in detail. The fault-masking ratio of the proposed voter (FMV) is 66% considering single/multiple faults. Simulation results establish the validation of the proposed logic in QCA, which targets nano-scale devices. The proposed logic is also suitable for conventional CMOS technology, which is verified with the Cadence tool.
  Keywords: quantum dot cellular automata; triple modular redundancy; fault-tolerant majority voter; QCA defects; reliability; nanoelectronics.
  
  
• Time series clustering using stochastic and deterministic influences  
  by Mirlei Silva, Rodrigo Mello, Ricardo Rios 
  Abstract: As part of the unsupervised machine learning area, time series clustering aims at designing methods to extract patterns from temporal data in order to organise different series according to their similarities. According to the literature, most of researches either perform a preprocessing step to convert time series into an attribute-value matrix to be later analyzed by traditional clustering methods, or apply measures specifically designed to compute the similarity among time series. Based on such studies, we have noticed two main issues: i) clustering methods do not take into account the stochastic and the deterministic influences inherent of time series from real-world scenarios; and ii) similarity measures tend to look for recurrent patterns, which may not be available in stochastic time series. In order to overcome such drawbacks, we present a new clustering approach that considers both influences and a new similarity measure to deal with purely stochastic time series. Experiments provided outstanding results, emphasizing time series are better clustered when their stochastic and deterministic influences are properly analysed.
  Keywords: time series; clustering; similarity measure.
  DOI: 10.1504/IJCSE.2019.10022008
   
  
• Laius: an energy-efficient FPGA CNN accelerator with the support of a fixed-point training framework  
  by Zikai Nie, Zhisheng Li, Lei Wang, Shasha Guo, Yu Deng, Rangyu Deng, Qiang Dou 
  Abstract: With the development of Convolutional Neural Networks (CNNs), their high computational complexity and energy consumption become significant problems. Many CNN inference accelerators have been proposed to reduce the energy consumption. Most of them are based on 32-bit float-point matrix multiplication, where the data precision is over-provisioned for inference. This paper presents Laius, an 8-bit fixed-point LeNet inference engine implemented on FPGA. To economise FPGA resources, we propose a methodology to find the optimal bit-length for weight and bias in LeNet. We use optimisations of pipelining, PE tiling, and theoretical analysis to improve the performance. Moreover, we optimise the convolutional sequence and data layout for further research. Experiment results show that Laius achieves 44.9 Gops throughput. Moreover, with only 1% accuracy loss, Laius largely reduces 31.43% in delay, 87.01% in LUT consumption, 66.50% in BRAM consumption, 65.11% in DSP consumption and 47.95% reduction in power compared with the 32-bit version with the same structure.
  Keywords: CNN accelerator; FPGA; inference engine; fixed-point training; data layout.
  
  
• A graph representation for search-based approaches to graph layout problems  
  by Behrooz Koohestani 
  Abstract: A graph consists of a finite set of vertices and edges. Graphs are used to represent a significant number of real life applications. For example, in computer science, graphs are employed for the representation of networks of communication, organisation of data, flow of computation, computational devices, etc. Several data structures have been proposed for representing graphs among which the adjacency matrix, adjacency list and edge list are the most important and widely used ones. The choice of a graph representation is mainly situation-specific and depends on the type of operations required to be performed on a given graph as well as the ease of use. In this research, a specialised graph representation is proposed, specifically designed for use when coping with graph-based optimisation problems (e.g., graph layout problems) through heuristic search methods with the aim of speeding up the search. The results of numerical experiments show that for the purpose of this study, the proposed approach performs extremely well compared to well-known graph representation approaches.
  Keywords: graph representation; combinatorial optimisation; graph layout problems; search methods.
  DOI: 10.1504/IJCSE.2019.10020775
   
  
• Energy-efficiency-aware flow-based access control in HetNets with renewable energy supply  
  by Li Li, Yifei Wei, Mei Song, Xiaojun Wang 
  Abstract: Software defined networking (SDN) is revolutionising the telecommunication networking industry by providing flexible and efficient management. This paper proposes an energy-efficiency-aware flow-based management framework for relay-assisted heterogeneous networks (HetNets), where the relay nodes are powered by renewable energy. Owing to the dynamic property of user behaviour and renewable energy availability, the flow-based management layer should enhance not only the instantaneous energy efficiency but also the long-term energy efficiency, while satisfying the transmission rate demand for each user. We first formulate the energy efficiency problem in HetNets as an optimisation problem for instantaneous energy efficiency and renewable energy allocation, and propose a heuristic algorithm to solve the optimisation problem. According to the proposed algorithm, we then design a dynamic flow-table configuration policy (DFTCP), which can be integrated as an application on top of an SDN controller to enhance the long-term energy efficiency. Simulation results show that the proposed policy can achieve higher energy efficiency compared with current distributed relay strategy, which chooses the nearest or strongest signal node to access, and obtain better performance for the overall relay network when the user density and demand change.
  Keywords: software defined networking; energy efficiency; renewable energy.
  
  
• MOEA for discovering Pareto-optimal process models: an experimental comparison  
  by Sonia Kundu, Manoj Agarwal, Shikha Gupta, Naveen Kumar 
  Abstract: Process mining aims at discovering the workflow of a process from the event logs that provide insights into organisational processes for improving these processes and their support systems. Process mining abstracts the complex real-life datasets into a well-structured form known as a process model. In an ideal scenario, a process mining algorithm should produce a model that is simple, precise, and general, and fits the available logs. A conventional process mining algorithm typically generates a single process model that may not describe the recorded behaviour effectively. Multi-objective evolutionary algorithms (MOEA) for process mining optimise two or more objectives to generate several competing process models from the event logs. Subsequently, a user can choose a model based on his/her preference. In this paper, we have experimentally compared the popular second-generation MOEA algorithms for process mining.
  Keywords: process discovery; evolutionary algorithms; Pareto front; multi-objective optimisation; process model quality dimensions; PAES; SPEA-II; NSGA-II.
  
  
• ElBench: a microbenchmark to evaluate virtual machine and container strategies on executing elastic applications in the cloud  
  by Rodrigo Da Rosa Righi, Cristiano Costa, Adenauer Yamin, Vinicius Facco, Douglas Brauner 
  Abstract: One of the main features of cloud computing that differentiates it from clusters and grids is the elasticity of resources, being mainly implemented through virtual machines (VMs) that deliver an easy mechanism for replication and isolation. In particular, in the high performance computing (HPC) panorama, the use of VMs to run parallel applications can impose prohibitive overheads, either in terms of time penalties related to scaling out operations or in terms of large delays on accessing hypervisor-based virtualised hardware. In addition to VMs, today we perceive the emergence of the container technology to implement the aforementioned facilities; however, our investigation did not discover an initiative that describes a comparison formalism to evaluate VM and container techniques to run HPC elastic applications in the cloud. This article explores this gap, presenting a microbenchmark named ElBench, which focuses on providing a framework to compare VM and container on executing elastic parallel applications in the cloud. Using a starting infrastructure and a predefined number of maximum and minimum resources, ElBench provides runtime traces along the execution, in addition to the conclusion time, resource use and cost (time
  Keywords: benchmark; cloud elasticity; HPC; container; virtual machine; virtualisation.
  DOI: 10.1504/IJCSE.2019.10021443
   
  
• Efficient web service selection with uncertain QoS  
  by Fethallah Hadjila, Amine Belabed, Mohammed Merzoug 
  Abstract: The QoS-based service selection in a highly dynamical environment is becoming a challenging issue. In practice, the QoS fluctuations of a service composition entail major difficulties in measuring the degree to which the user requirements are satisfied. In addition, the search space of feasible compositions (i.e., the solutions that preserve the requirements) is generally large and cannot be explored in a limited time; therefore, we need an approach that not only copes with the presence of uncertainty but also ensures a pertinent search with a reduced computational cost. To tackle this problem, we propose a constraint programming framework and a set of ranking heuristics that both reduce the search space and ensure a set of reliable compositions. The conducted experiments show that the ranking heuristics, termed 'fuzzy dominance' and 'probabilistic skyline', outperform almost all existing state-of-the-art methods.
  Keywords: web service selection; QoS uncertainty; global QoS conformance; constraint programming.
  
  
• PCR: caching replacement algorithm in proxy server  
  by Tong Liu, Xiaoyu Peng, Jiahao Liang, Jianhua Lu, Baili Zhang 
  Abstract: The efficiency of caching is a key factor affecting the performance of Content Delivery Network (CDN). The main aim of current CDN caching is to obtain a higher hit ratio, since the validation and freshness of outdated pages have not received due consideration in their replacement model. In this paper, a new improved cache profit model is clearly defined, and the freshness factors of web pages have been taken into adequate account. Based on the profit model, a new replacement algorithm-PCR (Proxy Cache Replacement) is recommended, and it can be proved optimal under the rational hypothesis. To conclude, a series of comparative experiments verified the efficiency of PCR in web caching replacement.
  Keywords: cache profit; replacement mechanism; proxy server.
  
  
• Prediction of gold-bearing localised occurrences from limited exploration data  
  by Igor Grigoryev, Adil Bagirov, Michael Tuck 
  Abstract: Inaccurate drill-core assay interpretation in the exploration stage presents challenges to long-term profit of gold mining operations. Predicting the gold distribution within a deposit as precisely as possible is one of the most important aspects of the methodologies employed to avoid problems associated with financial expectations. The prediction of the variability of gold using a very limited number of drill-core samples is a very challenging problem. This is often intractable using traditional statistical tools where, with less than complete spatial information, certain assumptions are made about gold distribution and mineralisation. The decision-support predictive modelling methodology based on the unsupervised machine learning technique presented in this paper avoids some of the restrictive limitations of traditional methods. It identifies promising exploration targets missed during exploration and recovers hidden spatial and physical characteristics of the explored deposit using information directly from drill hole database.
  Keywords: unsupervised machine learning; mathematical programming; resource definition; prediction; clusterwise linear regression.
  
  
• Mutual-inclusive learning-based multi-swarm PSO algorithm for image segmentation using an innovative objective function  
  by Rupak Chakraborty, Rama Sushil, Madan Garg 
  Abstract: This paper presents a novel image segmentation algorithm formed by the Normalised Index Value (Niv) and Probability (Pr) of pixel intensities. To reduce the com putational complexity, a mutual-inclusive learning-based optimisation strategy, named Mutual-Inclusive Multi-swarm Particle Swarm Optimization (MIMPSO) is also proposed. In mutual learning, a high dimensional problem of Particle Swarm Optimisation (PSO) is divided into several one-dimensional problems to get rid of the high dimensionality problem whereas premature convergence is removed by the inclusive-learning approach. The proposed Niv and Pr based technique with the MIMPSO algorithm is applied on the Berkley Dataset (BSDS300) images, which produce better optimal thresholds at a faster convergence rate with high functional values compared with the considered optimisation techniques such as PSO, Genetic Algorithm (GA) and Artificial Bee Colony (ABC). The overall performance in terms of the fidelity parameters of the proposed algorithm is carried out over the other stated global optimisers.
  Keywords: multilevel thresholding; normalised index value; probability; multi-swarm PSO.
  DOI: 10.1504/IJCSE.2020.10024788
   
  
• VFS_CS: a light-weight and extensible virtual file system middleware for cloud storage system  
  by Zeng Saifeng 
  Abstract: In cloud environments, data-intensive applications have been widely deployed to solve non-trivial tasks, while cloud-based storage systems usually fail to provide desirable performance and efficiency when running those data-intensive applications. To address the problems of storage capacity and performance when executing data-intensive applications, we design and implement a light-weight distributed file system middleware, namely Virtual File System for Cloud Storage, which allows other storage-level services to be easily incorporated into an integrated framework in a plug-in manner. In the proposed middleware, we implement three effective mechanisms: disk caching, file striping, and metadata management. The implementation of the proposed middleware has been deployed in a realistic cloud platform, and its performance has been thoroughly investigated under various workloads. Experimental results show that it can significantly improve I/O performance comparing with existing approaches. In addition, it also exhibits better robustness when the cloud system is facing intensive workloads.
  Keywords: cloud computing; distributed storage; file system; data layout; disk cache.
  
  
• Performance analysis of non-linear activation function in convolution neural network for image classification  
  by Edna Too, Li Yujian, Pius Kwao Gadosey, Sam Njuki, Firdaous Essaf 
  Abstract: Deep learning architectures which are exceptionally deep have exhibited to be incredibly powerful models for image processing and classification and general computer vision. As the architectures become deep, they introduce challenges and difficulties in the training process, such as overfitting, computational cost, and exploding/vanishing gradients and degradation. A new state-of-the-art densely connected architecture, DenseNets, has exhibited an exceptionally outstanding result for image classification. However, it is still computationally costly to train DenseNets. Several approaches have been recommended to deal with deeper network issues, including nonlinear activation functions. The choice of the activation function is also an important aspect in training of deep learning networks because it has a considerable impact on the training and performance of a network model. Therefore, an empirical analysis of some of the nonlinear activation functions used in deep learning is done for image classification and identification. The activation functions evaluated include ReLU, Leaky ReLU, ELU, SELU and an ensemble of SELU and ELU. Publicly available datasets Cifar-10, SVHN, and PlantVillage are used for evaluation. From the experimental results, SELU has a tendency to be more accurate and parameter efficient. Equally, it is seen to be fairly fast compared with ReLU and LeakyReLU. It achieves the testing accuracy score of 99.5%, 93.7% and 83.05% on PlantVillage, SVHN, and Cifar-10, respectively. Fast, accurate and parameter efficiency is desired to train DenseNets models.
  Keywords: deep learning; convolutional neural network; activation functions; nonlinear activation functions; image classification.
  
  
• User content categorisation model, a generic model that combine text mining and semantic models  
  by Randa Benkhelifa, Ismaïl Biskri, Fatima Zohra Laallam, Esma Aïmeur 
  Abstract: Social networking websites are growing not only regarding users number but also in the term of the user-generated content. These data represent a valuable source of information for several applications, which require the meaning of that content associated with the personal data. However, the current structure of social networks does not allow extracting in a fast and straightforward way the hidden information sought by these applications. Despite, major efforts have emerged from the semantic web community addressing this problem trying to represent the user as accurately as possible. They are not unable to give a sense to the user-generated content. For this, more sense-making need to be done on the content, to enrich the user profile. In this paper, we introduce a generic model called user content categorisation (UCC). It incorporates the text mining approach into a semantic model to enrich the user profile by including information on user's posts classifications.
  Keywords: semantic models; ontology; text mining; machine learning; user interests; user categorisation; text categorisation; profiling; ontology learning.
  
  
• Fine-tuning of pre-trained convolutional neural networks for diabetic retinopathy screening: a clinical study  
  by Saboora Mohammadian Roshan, Ali Karsaz, Amir Hossein Vejdani, Yaser Mohammadian Roshan 
  Abstract: Diabetic retinopathy is a serious complication of diabetes, and if not controlled, may cause blindness. Automated screening of diabetic retinopathy helps physicians to diagnose and control the disease in early stages. In this paper, two case studies are proposed, each on a different dataset. Firstly, automatic screening of diabetic retinopathy using pre-trained convolutional neural networks was employed on the Kaggle dataset. The reason for using pre-trained networks is to save time and resources during training compared with fully training a convolutional neural network. The proposed networks were fine-tuned for the pre-processed dataset, and the selectable parameters of the fine-tuning approach were optimised. At the end, the performance of the fine-tuned network was evaluated using a clinical dataset comprising 101 images. The clinical dataset is completely independent of the fine-tuning dataset and is taken by a different device with different image quality and size.
  Keywords: deep learning; convolutional neural network; diabetic retinopathy; inception model; clinical study.
  
  
• A deep neural architecture for sentence semantic matching  
  by Xu Zhang, Wenpeng Lu, Fangfang Li, Ruoyu Zhang, Jinyong Cheng 
  Abstract: Sentence semantic matching (SSM) is a fundamental research task in natural language processing. Most existing SSM methods take advantage of sentence representation learning to generate a single or multi-granularity semantic representation for sentence matching. However, sentence interactions and loss function, which are the two key factors for SSM, still haven't been fully considered. Accordingly, we propose a deep neural network architecture for SSM task with a sentence interactive matching layer and an optimised loss function. Given two input sentences, our model first encodes them to embeddings with an ordinary long short-term memory (LSTM) encoder. Then, the encoded embeddings are handled by an attention layer to find the key and important words in sentence. Next, sentence interactions are captured with a matching layer to output a matching vector. Finally, based on the matching vector, a fully connected multi-layer perceptron outputs the similarity score. The model also distinguishes the equivocation training instances with an improved optimised loss function. We also systematically evaluate our model on a public Chinese semantic matching corpus, BQ corpus. The experiment results demonstrate that our model outperforms the state-of-the-art methods, i.e., BiMPM and DIIN.
  Keywords: sentence matching; sentence interaction; loss function.
  
  
• Reversibly hiding data using dual images scheme based on EMD data hiding method  
  by Yu Chen, Jiangyi Lin, Chin-Chen Chang, Yu-Chen Hu 
  Abstract: This paper presents a novel grayscale image reversible data hiding scheme based on the exploiting modification direction (EMD) method. In this scheme, two 5-ary secret numbers are embedded into each pixel pair in the cover image according to the EMD method to generate two pairs of stego pixels. Two meaningful shadow images are obtained by shifting the generated corresponding pixel pairs, and the original image and the secret data can be accurately recovered when the two shadow images are operated together. Experimental results show that the proposed scheme has a good performance in the shadow image quality and the image embedding ratio.
  Keywords: reversible data hiding; secret image sharing; exploiting modification direction.
  
  
• The internet of things for healthcare: optimising e-health system availability in fog and cloud computing  
  by Guto Leoni Santos, Demis Gomes, Judith Kelner, Djamel Sadok, Francisco Airton Silva, Patricia Takako Endo, Theo Lynn 
  Abstract: The integration of fog and cloud computing has enabled a multitude of Internet of Things (IoT) applications through greater scalability, availability and connectivity. E-health systems can be used to monitor and assist people in real time, offering a range of multimedia-based health services, at the same time reducing the system cost since cheaper sensors and devices can be used to compose it. However, any downtime, mainly in the case of critical health services, can result in patient health problems and in the worst case, loss of life. In this paper, we use an interdisciplinary approach combining stochastic models with optimisation algorithms to analyse how failures impact e-health monitoring system availability. We propose stochastic-based surrogate models to estimate the availability of e-health monitoring systems that rely on edge, fog, and cloud infrastructures. Then, based on these surrogate models, we apply a multi-objective optimisation algorithm, NSGA-II, to improve system availability considering component costs as a constraint. Results suggest that replacing components with more reliable ones is more effective in improving the availability of an e-health monitoring system than adding more redundant components.
  Keywords: availability; cloud computing; edge computing; e-health systems; fog computing; internet of things; optimisation algorithms; stochastic models; surrogate models.
  
  
• A method of automatic text summarisation based on long short-term memory  
  by Wei Fang, Tianxiao Jiang, Ke Jiang, Yewen Ding, Feihong Zhang, Sheng Jack 
  Abstract: Deep learning is currently developing very fast in the NLP field and has achieved many amazing results in the past few years. Automatic text summarisation means that the abstract of the document is automatically summarised by a computer program without changing the original intention of the document. There are many application scenarios for automatic summarisation, such as news headline generation, scientific document abstract generation, search result segment generation, and product review summarisation. In the era of internet big data in the information explosion, if the short text can be used to express the main connotation of information, it will undoubtedly help to alleviate the problem of information overload. In this paper, a model based on a long short-term memory network is presented to automatically analyse and summarise Chinese articles by using the seq2seq + attention models. Finally, the experimental results are attached and evaluated.
  Keywords: deep learning; text summarisation; NLP; RNN; LSTM; seq2seq; attention; jieba; separate words; the language model.
  
  
• Broker-based mechanism for cloud provider selection  
  by Raghavendra Achar, P. Santhi Thilagam, Shreenath Acharya 
  Abstract: Cloud computing has emerged as a new paradigm for delivering on demand virtualised computing resources over the internet on a pay-per-use basis. Many cloud providers have started offering a variety of hardware configurations, operating systems and supporting services with varying pricing models to meet the increasing demands of different cloud customers. Despite the growing adoption and benefits of the cloud, it is still a challenging issue for cloud providers to adaptively manage the virtualised resources for diverse set of applications with unpredictable time-varying workloads while meeting cloud customers' Quality of Service (QoS) defined in terms of Service Level Agreements (SLAs). Applications hosted in the cloud have different QoS requirements, which include both low level (resource) requirements, such as CPU time, memory size, network bandwidth, and disk space, and high level (performance)requirements, such as security, availability, response time, and throughput. However, most cloud providers satisfy SLAs based on resource requirements rather than providing performance guarantees to applications. This gap creates a need for selecting a more suitable cloud provider who can satisfy performance requirements of applications along with resource requirements. This work aims at proposing a broker-based approach to rank cloud providers based on QoS requirements of customers and Key Performance Indicators (KPIs) of services offered by cloud providers. It helps the SaaS providers to save cost and complexity in choosing a suitable cloud provider for hosting applications. The experiment results show that proposed approach selects the suitable cloud provider for hosting various types of application satisfying the needs of different cloud customers.
  Keywords: cloud broker; selection; provider.
  
  
• Self-organised resource assignment for on-demand services in the cloud platform  
  by Zhiqiang Ruan, Dan Yang 
  Abstract: It is more popular for a multimedia service provider (MCSP) to deploy many data centers (DCs) in different geographic locations over cloud for delivering Video-on-Demand (VoD) services to a lot of users. One primary task of the MCSP is to maximise its profit while guaranteeing the users quality-of-service (QoS) requirements. However, the stochastic arrival of user request and the capacity restriction of individual DC make resource management in in a distributed cloud more challenging than in a general cloud. We present a resource assignment strategy that can accommodate heterogeneous network resources and QoS demands by converting the request distribution problem into the constrained function optimisation problem, an online algorithm is developed and certified approximating to the optimum solution. Compared with alternatives, our algorithm can cut down more than 35% operational cost without degrading the QoS of end users.
  Keywords: self-organised; resource assignment; cloud platform.
  
  
• Dynamic negotiation of user behaviour via blockchain technology in a federated system  
  by Min Yang, Shibin Zhang, Yang Zhao, Qirun Wang 
  Abstract: With the increasing number of network systems and users, a myriad of data are generated by users' daily activities, then comes the big data era. Instead of concentrating on how to realise information sharing, which is a hot topic in current research, the focus of this paper tries to establish a model that will realise user trust negotiation with the help of blockchain technology. As far as we know, blockchain technology has been widely employed in private information sharing, even in the application of shared economy, such as medical data sharing, smart grid data sharing as well as many new shared economy applications developed with a combination of Internet of Things (IoT) and blockchain. However, little attention has been paid to the new threat posed by user behaviour in such a heterogeneous environment; a malicious user may gather a huge amount of information concerning normal users from different network systems after registration and logging, and various attacks could be initiated by faulty users. Therefore, in this paper, how to not only detect a malicious user but also negotiate the users' trust value among network systems are discussed. The dominating work is as follows: firstly, a variety of decentralised systems are considered as nodes of P2P networks, and form a federated system with a certain number; secondly, every user behaviour profile is anchored on the user behaviour blockchain among the federated system, which concludes the trajectory of a user overall behaviours, on the basis of user behaviour profile, the trust negotiation model based on Practical Byzantine Fault Tolerance(PBFT) is proposed. Finally, a scenario based on the model is proposed and its safety problems are analysed, which makes a new try in dynamic negotiation of user trust.
  Keywords: federated system; user behaviour profile; blockchain technology; PBFT; dynamic negotiation.
  
  
• An influence maximisation algorithm based on community detection  
  by Yan Yuan, BoLun Chen, YongTao Yu, Ying Jin 
  Abstract: Influence maximisation is an important research direction in social networks. The main goal of this approach is to select seed nodes in the network to maximise the propagated influence. Because the influence maximisation is an NP-hard problem, existing studies have provided approximate solutions, and the research focuses on the framework of greed, but the time complexity of the greedy algorithm is high. In this study, an influence maximization algorithm based on community detection is proposed. This algorithm uses the K-means algorithm to divide the community. According to the modularity, the optimal community segmentation result is selected. By calculating the edgebetweenness of each community, some nodes are selected as important nodes. The important nodes of each community constitute the set of seed nodes used in the influence maximisation algorithm. Experiments show not only that the algorithm has an improved influence, but also that the time complexity is effectively reduced.
  Keywords: community detection; modularity; influence maximisation.
  
  
• Pipeline image haze removal system using dark channel prior on cloud processing platform  
  by Ce Li, Tan He, Yingheng Wang, Liguo Zhang, Ruili Liu, Jing Zheng 
  Abstract: Pipeline disease detection is a very important application of pipeline robots for the security of underground drainage pipeline facilities. The detection performance of existing systems is closely related to the image definition in the complex pipeline environment in terms of darkness, water fog, haze, etc. In this paper, the techniques of dark channel prior and cloud processing are combined into the framework of pipeline image haze removal system. In the system, including the user management module, system sitting module, cloud-based image management module and image processing module, we transmit the image data with the secure cloud data control mechanism, and remove the haze in each image using dark channel prior. The experimental results show that the system has good effects on haze removal of pipe images, especially for the larger reflection area. The system can be applied to engineering practice.
  Keywords: pipeline image processing; dark channel prior; atmospheric optical; data access control.
  
  
• Application research on service innovation and entrepreneurship education in university libraries and archives  
  by Xiufang Qian, Huamei Shi, Chunpeng Ge, Honghui Fan, Xiaorong Zhao, Yijun Liu 
  Abstract: College students' innovation and entrepreneurship education is an education to cultivate college students' innovative spirit, entrepreneurial awareness and ability to improve innovation and entrepreneurship. At present, the actual development effect of innovation and entrepreneurship education in China is not obvious. The innovation ability of college students needs to be improved, the proportion of innovation and entrepreneurship is still relatively low, and the success rate of entrepreneurship is not high. Therefore, we suggest that university libraries need to make full use of big data technology to better serve innovation and entrepreneurship education, thus improving the effectiveness of innovation and entrepreneurship education, and facilitating the development of more innovative and entrepreneurial talents. Firstly, this paper expounds the connotation of innovation and entrepreneurship education and analyses the application of big data in university libraries, and secondly it proposes the mechanism of cross-functional collaboration. Then, by integrating the independent data resources of university functional departments, the mechanism overcomes the problem of fragmentation; finally, it innovatively proposes the establishment of new archives, and then through the means of big data for innovation and entrepreneurship education, ultimately improves the quality of innovation and entrepreneurship education.
  Keywords: university library; archives; innovation and entrepreneurship education; intelligence information; personalised recommendation.
  
  
• Hierarchical routing protocol in wireless sensor networks: a state-of-the-art review  
  by Weidong Fang, Wuxiong Zhang, Lianhai Shan, Biruk Assefa, Wei Chen 
  Abstract: The hierarchical routing protocol in a wireless sensor network (WSN) is widely applied because of its good network stability and effective communication capability. Its cluster structure and the integration process of the data have a good performance in energy consumption and data transmission. To guide the design of hierarchical routing protocol, in this paper, we review the current research on these protocols, some typical technical characteristics such as the cluster head selection, clustering, and data routing is analysed, and the suggestion on the direction of the hierarchical routing protocol research is given. The work of this paper will be beneficial to the research, design, and optimisation of the hierarchical routing protocol.
  Keywords: wireless sensor network; hierarchical routing protocols; LEACH; PEGASIS; TEEN.
  
  
• A new encrypted image retrieval method based on feature fusion in cloud computing  
  by Jiaohua Qin 
  Abstract: With the rapid development of big data and internet security, content-based image retrieval has been widely studied and applied. Since image retrieval has great requirements on the computing power and storage capacity of the platform, the cloud server has become the preferred choice for outsourcing image retrieval. However, the cloud server is not completely reliable, and outsourcing image retrieval may bring many security and privacy problems. The image retrieval scheme based on privacy protection usually extracts single features in the current, which results in low retrieval accuracy. To solve these problems, we propose a new encrypted image retrieval method based on feature fusion in cloud computing. Firstly, we encrypt the RGB channels of the image by the encryption operator. Then, we design a feature extractor, which can extract the enhanced RGB(E-RGB) and HSV colour histogram feature. Finally, we upload the encrypted image and the feature extractor to the cloud directly, and the cloud server can obtain the encrypted images E-RGB and HSV colour histogram feature weights and the fusion feature vector by a feature extractor. The cloud server can calculate the similarity between images by directly comparing the Euclidean distance between two feature vectors. The experiments and security analysis show that the proposed image retrieval method has good security and accuracy.
  Keywords: searchable encryption; multi-level feature fusion; encrypted image retrieval; enhanced RGB.
  
  
• Research on network layout strategy of mobile opportunity perception in coal mines  
  by Jingzhao Li, Zhi Xu 
  Abstract: In view of the problems of large-scale application cost of existing underground coal mine monitoring systems, difficulty in expanding the network, and difficulty in achieving complete coverage of coal mines, it is proposed to apply the opportunity network technology to the underground coal mine environment and combine it with the existing wired network and wireless network, by using down-hole mobile devices and personnel to construct a sparse heterogeneous converged network in which the opportunistic network acts as the core. Based on a sufficient survey, a new method of "wired + wireless", "A class fixed node + B class fixed node + C class mobile node + D class mobile node" is proposed to construct a coal mine underground data acquisition and transmission method. The system structure and the composition, function, connection rule and historical information of the four types of nodes are analysed. On this basis, the prediction and correction of the interaction between the artificial mobile node and the fixed information in the next cycle are realised, and the dynamic planning of the interaction path of the nodes is further realised. The system is applied in the industrial experiment of Xutuan Coal Mine of Huaibei Mining Group, realising the real-time, dynamic and full coverage of the information of the underground area of the mine, enriching the theory and application of the mine network of perceived mines, and providing a powerful guarantee for internet of things, interaction of objects, intelligent perception, intelligent processing in the mine safety management.
  Keywords: coal mine underground; opportunity perception; mobile network; sensory nodes; information interaction.
  
  
• A novel high capacity turtle shell based data hiding with location table free  
  by Jiang-Yi Lin, Yu Chen, Chin-Chen Chang, Yu-Chen Hu 
  Abstract: In this paper, we design a novel turtle shell based data hiding scheme. The proposed scheme starts by constructing a magic matrix based on turtle shells. By using the magic matrix, four secret bits can be concealed in a pixel pair selected from the cover image. In our experiments, 2 bit per pixel (bpp) of embedded capacity can be achieved of the proposed scheme and the secret data can be correctly extracted without storing any overhead messages. In addition, the proposed scheme is superior to some state-of-the-art data hiding schemes subject to the peak signal-to-noise ratio measurement and it has good performance on resisting the statistical attack of PVD histogram.
  Keywords: data hiding; magic matrix; turtle shell; location table.
  
  
• An improved Otsu threshold segmentation algorithm  
  by Pei Yang, Wei Song, Xiaobing Zhao, Rui Zheng, Letu Qingge 
  Abstract: Image segmentation is widely used as a fundamental step for various image processing applications. This paper focuses on improving the famous image thresholding method named Otsus algorithm. Based on the fact that threshold acquired by Otsus algorithm tends to be closer to the class with larger intraclass variance when the foreground and background have large intraclass variance difference, an improved strategy is proposed to adjust the threshold bias. We analysed the relationship between pixel grayscale value and the change of cumulative pixel number, and selected the ratio of pixel gray level value to a certain cumulative pixel number as the adjusted threshold. Experiments using typical testing images were set up to verify the proposed method both quantitatively and qualitatively. Two widely used metrics named misclassification error (ME) and Dice similarity coefficient (DSC) are adopted for quantitative evaluation, and both quantitative and qualitative results indicated that the proposed algorithm could better segment the testing images and get competitive misclassification error and DSC values compared with Otsus method and its improved versions proposed by Hu (Hu et al., 2009) and Xu (Xu et al., 2011), and the time consumption of our method can be significantly reduced.
  Keywords: Otsu's algorithm; threshold segmentation; maximum interclass variance; single threshold segmentation.
  
  
• A Scale Space Model of Weighted Average CNN Ensemble for ASL Fingerspelling Recognition  
  by Neena Aloysius, Geetha M 
  Abstract: A sign language translator is a utilitarian in facilitating communication between the deaf community and the hearing majority. This paper proffers an innovative specialised convolutional neural network (CNN) model, Sign-Net, to recognise hand gesture signs by incorporating scale space theory into a deep learning framework. The proposed model is an ensemble of CNNs - a Low Resolution Network (LRN) and a High Resolution Network (HRN). This architecture allows the ensemble to work at different spatial resolutions and at varying depths of CNN. The Sign-Net model was assessed with static signs of American Sign Language - alphabets and digits. Since there exists no sign dataset for deep learning, the ensemble performance is evaluated on the synthetic dataset that we have collected for this task. Assessment of the synthetic dataset by Sign-Net reported an impressive accuracy of 74.5%, notably superior to the other existing models.
  Keywords: convolutional neural networks; sign language; fingerspelling; ensemble; vgg-16; classification; scale space; spatial resolution.
  
  
• Deep characteristics analysis on travel time of emergency traffic  
  by Jiao Yao, Yaxuan Dai, Yiling Ni, Jin Wang, Jing Zhao 
  Abstract: Owing to the rapid development of emergency rescue transportation cities and the frequent emergencies, the demand for emergency rescue is increasing sharply. How to select an emergency rescue route quickly and shorten the rescue travel time under the condition of limited urban road resources is of great significance. Based on the characteristics analysis of emergency rescue, this paper classifies priority levels of different emergency traffic, moreover, the travel times are analysed with three scenarios: (1) no encounter queuing at the intersection, (2) encounter queues with available lanes, (3) encounter queues without available lanes. A related case study shows that the path model in this paper can effectively shorten the travel time of emergency traffic in the route and improve its efficiency.
  Keywords: emergency rescue traffic; deep characteristics analysis; travel time.
  
  
• Impact of climate changes on manufacturing: Hodrick Prescott filtering and a partial least squares regression model  
  by Keyao Chen, Guizhi Wang, Jibo Chen, Shuai Yuan, Guo Wei 
  Abstract: In order to explore the impact of climate change on manufacturing outputs in Nanjing, China, this paper first adopts a polynomial function to retrieve trend values of manufacturing output, and then elaborates to manipulate the Hodrick Prescott (HP) filtering to isolate the parts of manufacturing outputs that are affected by climate factors. Subsequently, the paper attempts to construct a Partial Least Squares Regression (PLSR) model covering meteorological factors (e.g. average annual temperature, precipitation, sunshine hours and four quarters' average temperatures) and manufacturing meteorological outputs. The results show that an increased average temperature and increased average precipitation yield negative impacts on manufacturing and production; in contrast, in winter, higher temperature offers benefits to manufacturing. Finally, this paper studies the changes of manufacturing outputs in Nanjing for different climate scenarios.
  Keywords: climate output; HP filter; multicollinearity; partial least squares regression.
  
  
• Intrusion detection of wireless sensor networks based on IPSO algorithm and BP neural network  
  by Xue Lu, Dezhi Han, Letian Duan, Qiuting Tian 
  Abstract: When the sensor nodes of wireless sensor networks (WSNs) are deployed to an open and unsupervised region, and they are vulnerable to various types of attack. Intrusion detection system can detect network attacks that nodes suffer from. This paper combines improved particle swarm optimisation (IPSO) algorithm and back-propagation neural network (BPNN), named IPSO-BPNN. We propose an intrusion detection model of WSNs based on a hierarchical structure. First, we use IPSO algorithm to optimise the initial parameters of BPNN to avoid falling into the local optimum. Then, we apply IPSO-BPNN to the intrusion detection of WSNs. Finally, we use benchmark NSL-KDD and UNSW-NB15 datasets to verify the performance of the IPSO-BPNN. The simulation results show that IPSO-BPNN has faster convergence speed, higher detection accuracy rate, and lower false positive rate compared with BPNN and BPNN optimised by PSO algorithm, which can meet the WSN intrusion detection requirements.
  Keywords: wireless sensor network; particle swarm optimisation; back-propagation neural network; intrusion detection.
  
  
• MLIM-Cloud: a flexible information monitoring middleware in large-scale cloud environments  
  by Tienan Zhang 
  Abstract: In large-scale cloud platforms, information monitoring service is essential for capturing the performance of underlying resources and understanding the behaviour of various applications in different circumstances. In this paper, we present a flexible information monitoring middleware, namely Multi-Level Information Monitoring for Cloud (MLIM-Cloud), which is designed to run in a non-intrusive and transparent manner in any virtualised infrastructure. In the MLIM-Cloud framework, three kinds of monitoring entity are designed for collecting, processing and achieving various kinds of runtime information at different infrastructure levels, including physical machines, VM instances, and up-level applications. In addition, the MLIM-Cloud middleware is both platform-independent and platform-interoperable, which means it can be easily deployed on different kinds of cloud platforms. To investigate the performance of MLIM-Cloud, an extensive set of experiments are conducted in a real-world cloud platform. The experimental results show that comparing with many existing monitoring services, the MLIM-Cloud middleware exhibits better adaptiveness and robustness when the cloud system is in presence of dynamic and unpredicted workloads.
  Keywords: cloud computing; information monitoring service; virtual machine; information filter; resource allocation.
  
  
• A multi-group e-commerce signature scheme based on quantum teleportation  
  by Jinqiao Dai, Shibin Zhang, Jinyue Xia 
  Abstract: In this paper, we propose a multi-group e-commerce signature scheme based on quantum teleportation and Bell states. Compared with the recent quantum group signature scheme, what our scheme has optimised is the traditional signature model with only one group manager and the use of quantum resources. By setting up different group managers in different groups, the scheme promotes the traditional signature model into a more practical situation and can be easily used in e-payment systems. In addition, the quantum teleportation and Pauli operations based on the Bell states are used in the transaction process, and these techniques can also be easily implemented under practical situations. Finally, the security analysis shows that the scheme can defend against different internal and external attacks, including intercepting resend attacks and entanglement attacks. The scheme is shown to be reliable, safe and effective when facing denial and forgery.
  Keywords: multi-group signature; e-commerce; Bell states; quantum teleportation.
  
  
• Numerical solution and Taguchi experimental method for variable viscosity and non-Newtonian fluids effects on heat and mass transfer by natural convection in porous media  
  by Ken Ming Tu, Kuo Ann Yih, Jyh Horng Chou 
  Abstract: In this article, both numerical solution and Taguchi method are presented to study the variable viscosity and non-Newtonian fluids effects on coupled heat and mass transfer by free convection over a vertical permeable plate in porous media. The surface temperature, concentration of the plate, and blowing/suction velocity are uniform. The viscosity of the fluid varies inversely as a linear function of the temperature. The partial differential equations are transformed into non-similar equations and solved by Keller box method. Numerical results of the local Nusselt number and local Sherwood number are expressed in the five parameters: 1) blowing/suction parameter ; 2) the power-law index of non-Newtonian fluid n; 3) buoyancy ratio N; 4) Lewis number Le; 5) viscosity-variation parameter r. The best value for confirming the maximum of the local Nusselt (Sherwood) number by the Taguchi method is 6.6328 (10.3056).
  Keywords: Taguchi experimental method; variable viscosity; non-Newtonian fluid; free convection; vertical permeable plate; porous media.
  
  
• Case data-mining analysis for patients with oesophageal cancer  
  by Yanning Cao, Xiaoshu Zhang, Jin Wang 
  Abstract: We are in an era of digital medicine in which physicians can generate copious patient data, but tools to analyse these data are limited. Thus, we used case data from patients with oesophageal cancer from a medical institution, removed incomplete information, and quantified the textual data according to recommendations from the corresponding physicians. We used different classification algorithms to process the data, predict patient survival, and compare accuracies across algorithms. Our experimental results show that the BayesNet algorithm was highly accurate and precise, and, thus, may represent a promising data-mining tool.
  Keywords: data mining; classification algorithms; oesophygeal cancer; BayesNet.
  
  
• Plaintext-aware encryption in the standard model under the linear Diffie-Hellman knowledge assumption  
  by Dongwei Gao, Hefeng Chen, Chin-Chen Chang 
  Abstract: In this paper, we consider the problem of constructing new plaintext-aware encryption in the standard model. A new hybrid asymmetric encryption scheme is presented using a new key encapsulation method and a data encapsulation method. To prove the presented asymmetric encryption scheme is simulatable, we put forward some new sufficient conditions for judging a group to be simulatable. By introducing a new assumption called linear Diffie-Hellman knowledge assumption, we prove the proposed hybrid asymmetric encryption scheme is PA2.
  Keywords: plaintext aware; PA2; linear Diffie-Hellman knowledge assumption; key encapsulation method; data encapsulation method.
  
  
• Collaborative humanless model for automatic pothole detection and driver notification  
  by Thiago Lopes, Cristiano Costa, Igor Fontana, Lucas Pfeiffer, Rodrigo Da Rosa Righi 
  Abstract: The bad conditions of roads in emerging countries characterised by potholes increase the occurrence of accidents, which sometimes also result in the loss of human lives. In this way, the thematic of identifying the location of the potholes is gaining more and more relevance, also helping drivers to plan better traffic routes. Both smartphone-assisted and collaborative applications appear as alternatives to mitigate such a problem, but they rely on the necessity of the user to be the trigger of the pothole identification and sharing procedures. In this context, this article presents the CoMDAP (Collaborative Model for Detection and Alert of Potholes), which provides a distributed framework that automatically collects, analyses and shares pothole and traffic data among users and drivers without any human interaction. In other words, our differential idea consists of using particular hardware in the vehicles to automatically detect the potholes in the roads, so increasing both the adoption of the model and the precise pothole location on each lane. The evaluation methodology first considers a prototype executed in simulated (a toy and in-home lanes) and real (a car in a particular road) scenarios in order to observe the accuracy of detecting the potholes. Second, also exploring the second scenario, we implement an Android application that notifies the drivers as they approach a pothole. The results were encouraging in both cases, highlighting the benefits of using CoMDAP as a counterpart to enable smart cities.
  Keywords: pothole detection; machine-to-machine collaboration; collaborative systems; safe driving; CoMDAP.
  
  
• An open speech resource for Tibetan multi-dialect and multitask recognition  
  by Yue Zhao, Xiaona Xu, Jianjian Yue, Wei Song, Xiali Li, Licheng Wu, Qiang Ji 
  Abstract: This paper introduces a Tibetan multi-dialect data resource for multitask speech research. It can be used for Tibetan multi-dialect speech recognition, Tibetan speaker recognition, Tibetan dialect identification, and Tibetan speech synthesis. The resource consists of 30 hours Lhasa-
  Keywords: Tibetan language; multi-dialect speech recognition; multitask learning; speech corpus.
  
  
• Transliteration recognition of Tibetan person name based on Tibetan cultural knowledge  
  by Zhijuan Wang, Wenguang Fang, Yinghui Feng, Xiaobing Zhao, Wei Song, Yining Chang 
  Abstract: ICTCLAS (Institute of Computing Technology, Chinese Lexical Analysis System) is a common tool for Chinese word segmentation and named entity recognition. With this tool, the F1 value of person name recognition from Chinese texts in Tibetan culture is only 40%. We proposed a method for trans-literation recognition of Tibetan person name based on Tibetan cultural knowledge. Firstly, we leveraged Tibetan cultural dictionary to improve the word segmentation performance of ICTCLAS. Then, special contextual features and naming rules of Tibetan person name in their culture were adopted to determine the boundary of Chinese transliteration of Tibetan person name. Finally, the transliteration candidates of Tibetan person name are filtered based on discrimination and reliability. Experiments on a 1.2M Tibetan text in Tibetan culture show that the method can increase the F1 value of Chinese transliteration recognition of Tibetan person name from 40.08% to 87.92% in ICTCLAS.
  Keywords: transliteration; Tibetan person name; Tibetan cultural knowledge; discrimination; reliability.
  
  
• Unambiguous discrimination of binary coherent states  
  by Wenbin Yu 
  Abstract: Considering quantum detection, there are two different ways to make a measurement on a signals set. One is minimum error discrimination and the other is unambiguous states discrimination. In this work, we study the unambiguous state discrimination of coherent states. The quantum measurement model investigated is based on binary signals. All necessary positive-operator valued measurements are established to implement the quantum measurements in non-ambiguous way. The conclusive probability and inconclusive probability for the unambiguous discrimination of both on-off keying and binary-phase-shifting keying modulations are derived rigorously to show the measurement performance of proposed detection method theoretically.
  Keywords: unambiguous state discrimination; quantum information processing; coherent state; quantum communication.
  
  
• A coverless information hiding algorithm based on gradient matrix  
  by Jianbin Wu, Chuwei Luo 
  Abstract: A coverless information hiding algorithm based on gradient matrix is introduced in this paper. Firstly, we grid the sharpened gradient matrix of an image. Then we encode the spectral radius of the gridded matrix to construct the mapping relationship between the matrix eigenvalues and random numbers. In order to improve the efficiency and security of transmission, we process the information segment by BCH (31,21) encoder, in which redundancy check bits are added to detect and recover the errors caused by the interference in communication, thus breaking the limit that the image library has to be shared between the sender and receiver. In the meantime, splicing strategy is adopted to reduce the difficulty of building image library with the length of information sequence increasing. As the experimental results show, this proposed algorithm has a strong robustness towards glitch attack, JPEG compression attack, etc. and has a great application value in high-level secret key communication.
  Keywords: coverless information hiding; gradient matrix; BCH coding; spectral radius; stitching strategy; robustness.
  DOI: 10.1504/IJCSE.2020.10024818
   
  
• An access model under cloud computing environment  
  by Wen Gu, Yang Cao, Yi Ying 
  Abstract: Concepts such as virtualisation, elasticity, and multi-tenancy have been embedded in cloud computing environments. Thus, the traditional access control model is no longer applicable to cloud computing environments, and designing a new access control model specific to the features of cloud computing environments is necessary. The software as a service (SaaS) pattern has gradually emerged as a type of cloud computing model that can address the information management requirements of small- and medium-sized enterprises. Given the features of SaaS application platforms, this study proposes a multi-tenant access control model called ST-RBAC under the SaaS pattern on the basis of a discussion of the current situation of access control models. The proposed model successfully deals with relationships among elements such as tenants, users, roles, and permissions. Hence, it can effectively guarantee user data safety and user permission management.
  Keywords: cloud computing; SaaS; multi-tenant; access control.
  
  
• Face spoof detection using feature map superposition and convolution neural network  
  by Fei Gu, Zhihua Xia, Jianwei Fei, Chengsheng Yuan, Qiang Zhang 
  Abstract: Face biometrics have been widely applied for user authentication systems in many practical scenarios, but the security of these systems can be jeopardised by presenting photos or replays of the legitimate user. To deal with such threats, many handcraft features extracted from face images or videos were used to detect spoof faces. These methods mainly analysed either illumination differences, colour differences or textures differences, but did not fuse these features together to further improve detection performance. Thus in this paper, we propose a novel face spoof detection method based on various feature maps and convolution neural network for photo and replay attacks. Specifically, both facial contour and specularly reflected features are considered, and proposed network is task-oriented designed, e.g. its depth and width, and specific convolutional parameters of each layer are chosen for optimal accuracy and efficiency. A remarkable performance through plenty of experiments on multiple datasets shows that our method can defend not only photo attack but also replay attack with a very low error probability.
  Keywords: face spoof detection; convolution neural network; difference of Gaussians; specular reflected light.
  
  
• Advances in the enumeration of foldable self-avoiding walks  
  by Christophe Guyeux, Jean-Claude Charr, Jacques Bou Abdo, Jacques Demerjian 
  Abstract: Self-Avoiding Walks (SAWs) have been studied for a long time owing to their intrinsic importance and the many application fields in which they operate. A new subset of SAWs, called foldable SAWs, has recently been discovered when investigating two different SAW manipulations embedded within existing Protein Structure Prediction (PSP) software. Since then, several attempts have been made to find out more about these walks, including counting them. However, calculating the number of foldable SAWs appeared as a tough work, and current supercomputers fail to count foldable SAWs of length exceeding ~30 steps. In this article, we present new progress in this enumeration, both theoretical (mathematics) and practical (computer science). A lower bound for the number of foldable SAWs is firstly proposed, by studying a special subset called prudent SAWs that is better known. The triangular and hexagonal lattices are then investigated for the first time, leading to new results about the enumeration of foldable SAWs on such lattices. Finally, a parallel genetic algorithm has been designed to discover new non-foldable SAWs of lengths ~100 steps, and the results obtained with this algorithm are promising.
  Keywords: self-avoiding walks; foldable SAWs; prudent SAWs; genetic algorithm.
  
  
• Distributed nested streamed models of tsunami waves  
  by Kensaku Hayashi, Alexander Vazhenin, Andrey Marchuk 
  Abstract: This research focuses on designing a high-speed scheme for tsunami modelling using nested computing. Computations are carried out on a sequence of grids composed of geographical areas with resolutions where each is embedded within another. This decreases the total number of calculations by excluding unimportant coastal areas from the process. The paper describes the distributed streaming computational scheme allowing for flexible reconfiguration of heterogeneous computing resources with a variable set of modelling zones. Computations are implemented by distributing these areas over modelling components and by synchronising the transitions of boundary data between them. Results of numerical modelling experiments are also presented.
  Keywords: tsunami modelling; nested grids; distributed systems; coarse-grained parallelisation; streaming computing; communicating processes; process synchronisation; task parallelism; programming model; component-based software engineering.
  
  
• E-commerce satisfaction based on synthetic evaluation theory and neural networks  
  by Jiayin Zhao, Yong Lu, H.A.O. Ban, Ying Chen 
  Abstract: The rapid development of e-commerce has led to the increasing role of satisfaction in more fields. Therefore, the customers opinion has become a necessary role for the success of related companies. E-commerce satisfaction, as the key factor affecting the performance of e-commerce enterprises, has become a research hotspot in academia. This paper proposes a synthetic evaluation model of satisfaction and logistics performance based on a fuzzy synthetic model and a dynamic weighted synthetic model, respectively. A modified ASCI analysis method based on the structured equation model is also proposed to compare with the synthetic method. Beyond this, we have also evaluated consumer satisfaction based review data. Corresponding suggestions are given for the operation of e-commerce enterprises.
  Keywords: E-commerce satisfaction; fuzzy synthetic model; structured equation model; neural networks.
  
  
• On the build and application of bank customer churn warning model  
  by Wangdong Jiang, Yushan Luo, Ying Cao, Guang Sun, Chunhong Gong 
  Abstract: In view of the customer churn problem faced by banks, this paper will use the Python language to clean and select the original dataset based on real bank customer data, and gradually condense the 626 customer features in the original dataset to 77 customer features. Then, based on the pre-processed bank data, this paper uses logistic regression, decision tree and neural network to establish three bank customer churn warning models and compares them. The results show that the accuracy of the three models in predicting bank loss customers is above 92%. Finally, based on the logistic regression model with better evaluation results, this paper analyses the characteristics of the lost customers for the bank, and gives the bank management suggestions for the lost customers.
  Keywords: bank customer; churn warning model; logistic regression; customer churn.
  
  
• Efficient deep convolutional model compression with an active stepwise pruning approach  
  by Shengsheng Wang, Chunshang Xing, Dong Liu 
  Abstract: Deep models are structurally tremendous and complex, thus making them hard to deploy on embedded hardware with restricted memory and computing power. Although the existing compression methods have pruned the deep models effectively, some issues exist in those methods, such as multiple iterations needed in the fine-tuning phase, difficulty in pruning granularity control, and numerous hyperparameters needed to set. In this paper, we propose an active stepwise pruning method of a logarithmic function which only needs to set three hyperparameters and a few epochs. We also propose a recovery strategy to repair the incorrect pruning thus ensuring the prediction accuracy of model. Pruning and repairing alternately constitute a cyclic process along with updating the weights in layers. Our method can prune the parameters of MobileNet, AlexNet, VGG-16 and ZFNet by a factor of 5.6
  Keywords: deep convolutional model; model compression; active stepwise pruning; parameter repairing; pruning intensity; logarithmic function.
  
  
• Forecasting yield curve of Chinese corporate bonds  
  by Maojun Zhang 
  Abstract: Forecasting the yield curve of corporate bonds is an important issue about the corporate bond pricing and its risk management. In this paper, the dynamic Nelson-Siegel model is used to fit the yield of the corporate bonds in China, and the AR model is used to forecast the yield curve. It is found that the Nelson-Siegel model fitting the yield of the corporate bonds with different credit ratings is notonly very effective but also can indicate the long-term, medium-term and short-term dynamic features of the yield curve. Moreover, the linear AR (1) model might be more suitable than the nonlinear AR(1) model.
  Keywords: corporate bonds; yield curve; Nelson-Siegel model; AR(1) model.
  
  
• Review on blockchain technology and its application to the simple analysis of intellectual property protection  
  by Wei Chen, Kun Zhou, Weidong Fang, Ke Wang, Fangming Bi, Biruk Assefa 
  Abstract: Blockchain is a widely used decentralised infrastructure. Blockchain technology has the decentration of network, the unforgeability of block data, etc. Therefore, blockchain technology has developed rapidly in recent years, and many organisations are involved. Applications are generally optimistic. This paper systematically introduces the background development status of blockchain, and analyses the operation mechanism, characteristics and possible application scenarios of blockchain technology from a technical perspective. Finally, the blockchain technology is applied to the intellectual property protection method as an example to study domestic and foreign examples and analyse existing problems. The review article aims to provide assistance for the application of blockchain technology.
  Keywords: blockchain; bitcoin; operating mechanism; intellectual property.
  
  
• An improved Sudoku-based data hiding scheme using greedy method  
  by Chin-Chen Chang, Guo-Dong Su, Chia-Chen Lin 
  Abstract: Inspired by Chang et al.s scheme, an improved Sudoku-based data hiding scheme is proposed here. The major idea of our improved scheme is to find the approximate optimal solution of Sudoku using the greedy method instead of through a brute-force search for an optimal solution. Later, the found approximate optimal solution of Sudoku is used to offer satisfactory visual stego-image quality with a lower execution time during the embedding procedure. Simulation results confirmed that the average stego-image quality is enhanced by around 90.51% compared with Hong et al.s scheme, with relatively less execution time compared with a brute-force search method.
  Keywords: data hiding; Sudoku; greedy method; brute-force search method; approximate optimal solution.
  
  
• A novel domain adaption approach for neural machine translation  
  by Jin Liu, Xiaohu Tian, Jin Wang, Arun Kumar Sangaiah 
  Abstract: Neural machine translation has been widely adopted in modern machine translation as it brings state-of-the-art performance to large-scale parallel corpora. For real-world applications, high-quality translation for text in a specific domain is crucial. However, the performance of general neural machine models drops when they are applied in a specific domain. To alleviate this issue, this paper presents a novel method of machine translation, which explores both model fusion algorithm and logarithmic linear interpolation. The method we propose can improve the performance of the in-domain translation model, while preserving or even improving the performance of the out-domain translation model. This paper has carried out extensive experiments on the proposed translation model using the public United Nations corpus. The BLEU (Bilingual Evaluation Understudy) score of the in-domain corpus and the out-domain corpus reaches 30.27 and 43.17, respectively, which shows a certain improvement over existing methods.
  Keywords: neural machine translation; model fusion; domain adaption.
  
  
• Formation path of customer engagement in virtual brand community based on back propagation neural network algorithm  
  by Lin Qiao, Mengmeng Song, Rob Law 
  Abstract: The formation path of customer engagement in a virtual brand community with customer engagement, which explores customers non-transactional behavior, has become increasingly popular in the marketing field. This paper introduces an approach that integrates structural equation modelling and back propagation artificial neural network to identify the motivating factors (interactivity, information quality, and convenience) that influence the perceived information value and social value of a virtual brand community and customer engagement. Our experiment shows that when perceived value plays a mediation role in the influence of interactivity and information quality on customer engagement, interactivity is positively associated with customer engagement in the virtual brand community. This study aims to provide meaningful implications to companies effective use of brand fan pages.
  Keywords: neural network algorithm; virtual brand community; brand fan page; perceived value.
  
  
• The mining method of trigger words for food nutrition matching  
  by Shunxiang Zhang 
  Abstract: The rational food nutrition matching plays a dual role in health and diet for humans. The trigger words related to food nutrition matching have an effect on classifying food nutrition matching into two types: reasonable nutrition matching and unreasonable nutrition matching. This paper proposes an aiming method of trigger words for food nutrition matching. First, food information frequency vector can be extracted by the number of food names, the number of nutrition ingredients and the number of matching effects in the sentence. By judging whether each component of food information frequency vector is 0 or not, the sentences unrelated to food nutrition matching can be filtered. Then, the two food verb-noun joint probability matrices can be constructed. The column of the first matrix is the food name, and the row is the verb. The column of the second matrix is the nutrition ingredient and matching effect, and the row is the verb. By comparing row mean value of the two matrices, whether the verb is a trigger word can be judged. Lastly, under the premise of commendatory and derogatory probabilities of the trigger word, the food nutrition matching can be classified as two types by naive Bayes. The experiments show that the proposed method effectively detects the trigger word related to food nutrition matching.
  Keywords: food nutrition matching; food information frequency vector; food verb-noun joint probability matrix.
  
  
• Estimating capacity-oriented availability in cloud systems  
  by Jamilson Dantas, Rubens Matos, Jean Teixeira, Eltton Túlio, Paulo Maciel 
  Abstract: Over the years, many companies have employed cloud computing to support their services and optimise their infrastructure usage. The provisioning of high availability and high processing capacity is a significant challenge when planning a cloud computing infrastructure. Even when the system is available, a part of the resources may not be offered owing to partial failures in just a few of the many components in an IaaS cloud. The dynamic behaviour of virtualised resources requires special attention to the effective amount of capacity that is available to users, so the system can be correctly sized. Therefore, the estimation of capacity-oriented availability (COA) is an important activity for cloud infrastructure providers to analyse the cost-benefit tradeoff among distinct architectures and deployment sizes. This paper presents a strategy to evaluate the COA of virtual machines in a private cloud infrastructure. The proposed strategy aims to provide an efficient and accurate computation of COA, by means of closed-form equations. We compare our approach with the use of models such as continuous time Markov chains, considering execution time and values of metrics obtained with both approaches.
  Keywords: capacity-oriented availability; closed-form equation; cloud computing; continuous time Markov chain.
  
  
• DDoS attack detection method based on network abnormal behaviour in big data environment  
  by Jing Chen, Xiangyan Tang, Jieren Cheng, Fengkai Wang, Ruomeng Xu 
  Abstract: Distributed denial of service (DDoS) attack is a rapidly growing problem with the fast development of the internet. The existing DDoS attack detection methods have time-delay and low detection rate. This paper presents a DDoS attack detection method based on network abnormal behaviour in a big data environment. Based on the characteristics of flood attack, the method filters the network flows to leave only the many-to-one network flows to reduce the interference from normal network flows and improve the detection accuracy. We define the network abnormal feature value (NAFV) to reflect the state changes of the old and new IP addresses of many-to-one network flows. Finally, the DDoS attack detection method based on NAFV real-time series is built to identify the abnormal network flow states caused by DDoS attacks. The experiments show that compared with similar methods, this method has higher detection rate, lower false alarm rate and lower missing rate.
  Keywords: DDoS; time series; ARIMA; big data; forecast.
  
  
• Two-level parallel CPU/GPU-based genetic algorithm for association rule mining  
  by Leila Hamdad, Zakaria Ournani, Karima Benatchba, Ahcène Bendjoudi 
  Abstract: Genetic algorithms (GA) are widely used in the literature to extract interesting association rules. However, they are time consuming mainly owing to the growing size of databases. To speed up this process, we propose two parallel GAs (ARM-GPU and ARM-CPU/GPU). In ARM-GPU, parallelism is used to compute the fitness which is the most time-consuming task; ARM-CPU/GPU proposes a two-level-based parallel GA. In the first level, the different cores of the CPU execute a GA-ARM on a sub-population. The second level of parallelism is used to compute the fitness, in parallel, on GPU. To validate the proposed two parallel GAs, several tests were conducted to solve well-known large ARM instances. Obtained results show that our parallel algorithms outperform state-of-the-art exact algorithms (APRIORI and FP-GROWTH) and approximate algorithms (SE-GPU and ME-GPU) in terms of execution time.
  Keywords: association rules; parallel genetic algorithm; GPU; CPU.
  
  
• A parallel adaptive-resolution hydraulic flood inundation model for flood hazard mapping  
  by Wencong Lai, Abdul Khan 
  Abstract: There is a growing demand for improved high-resolution flood inundation modelling in large-scale watersheds for sustainable planning and management. In this work, a parallel adaptive-resolution hydraulic flood inundation model is proposed for large-scale unregulated rivers. This model used the public best available topographic data and streamflow statistics data from USGS. An adaptive triangular mesh is generated with fine resolution (~30 m) around streams and coarse resolution (~200 m) away from streams. The river flood-peak discharges are estimated using the regression equations from the National Streamflow Statistics (NSS) Program based on watershed and climate characteristics. The hydraulic simulation is performed using a discontinuous Galerkin solver for the 2D shallow-water flow equations. The hydraulic model is run in parallel with the global domain partitioned using the stream link and stream length. The proposed model is used to predict the flooding in the Muskingum River Basin and the Kentucky River Basin. The simulated inundation maps are compared with FEMA maps and evaluated using three statistical indices. The results demonstrated that the model is capable of predicting flooding maps for large-scale unregulated rivers with acceptable accuracy.
  Keywords: flood inundation; flood mapping; unregulated rivers.
  
  
• The clothing image classification algorithm based on the improved Xception model  
  by Zhuoyi Tan, Yuping Hu, Dongjun Luo, Man Hu, Kaihang Liu 
  Abstract: This paper proposes a clothing image classification algorithm based on the improved Xception model. Firstly, the last fully connected layer of the original network is replaced with another fully connected layer to recognise eight classes instead of 1000 classes. Secondly, the activation function we employ in our network adopts both Exponential Linear Unit (ELU) and Rectified Linear Unit (ReLU), which can improve the nonlinear and learning characteristics of the networks. Thirdly, in order to enhance the anti-disturbance capability of the network we employ the L2 regularisation method. Fourthly, we perform data augmentation on the training images to reduce over-fitting. Finally, the learning rate is set to zero in the layers of the first two modules of our network and the network is fine-tuned. The experimental results show that the top-1 accuracy by the algorithm proposed in this paper is 92.19%, which is better than the state-of-the-art models of Inception-v3, Inception-ResNet-v2 and Xception.
  Keywords: clothing image classification; transfer learning; deep convolutional neural network; Xception.
  
  
• Research on image of enterprise after-sales service based on text sentiment analysis  
  by Yonghui Dai, Ying Wang, Bo Xu, Yingyi Wu, Jin Xian 
  Abstract: In recent years, the popularity of the internet has not only brought convenience to consumers, but also brought opportunities and challenges to enterprises. Among them, online reviews have a great impact on the enterprises, especially consumer reviews of the enterprises after-sales service will affect the enterprises image. In this paper, text sentiment analysis method was used to the analysis of after-sale online comments. According to the analysis results, enterprises can find the shortcomings of after-sales service and improve it. This paper provides the steps of the text sentiment analysis method, and uses the empirical data of the website to carry out the experiment. The results show that the method can effectively analyse the customer's sentiment and help the after-sales staff of the company to answer questions well, thereby improving the level of after-sales service and enterprise image.
  Keywords: enterprise image; after-sales service; sentiment analysis; online comment.
  
  
• A hidden Markov model to characterise motivation level in MOOCs Learning  
  by Yuan Chen, Dongmei Han, Lihua Xia, Jin Xian 
  Abstract: A participants learning in massive open online courses (MOOCs) highly relies on motivation. However, how to characterise motivation level is an open question. This study establishes a hidden Markov model to characterise motivation level and examines the model on the data from MOOCs in China. The empirical results show that two motivation levels (high and low) are characterised. Based on the two motivation levels, further analysis reveals the differences in participants learning behaviours with respect to learning activity and continuous learning. The hidden Markov model proposed in this study contributes to the development of the theoretical ground of current MOOCs. It also has important operational implications for MOOCs.
  Keywords: massive open online courses; motivation Level; hidden Markov model; learning behaviours.
  
  
• A short text conversation generation model combining BERT and context attention mechanism  
  by Huan Zhao, Jian Lu, Jie Cao 
  Abstract: The standard Seq2Seq neural network model tends to generate general and safe responses (e.g., I dont know) regardless of the input in the field of short-text conversation generation. To address this problem, we propose a novel model that combines the standard Seq2Seq model with the BERT module (a pre-trained model) to improve the quality of responses. Specifically, the encoder of the model is divided into two parts: one is the standard seq2seq which generates a context attention vector; the other is the improved BERT module which encodes the input sentence into a semantic vector. Then through a fusion unit, the vectors generated by the two parts are fused to generate a new attention vector. Finally, the new attention vector is transmitted to the decoder. In particular, we describe two ways to acquire a new attention vector in the fusion unit. Empirical results from automatic and human evaluations demonstrate that our model improves the quality and diversity of the responses significantly.
  Keywords: Seq2Seq; short text conversation generation; BERT; attention mechanism; fusion unit.
  
  
• Implicit emotional tendency recognition based on disconnected recurrent neural networks  
  by Yiting Yan, Zhenghong Xiao, Zhenyu Xuan, Yangjia Ou 
  Abstract: Implicit sentiment orientation recognition classifies emotions. The development of the internet has diversified the information presented by text data. In most cases, text information is positive, negative, or neutral. However, the inaccurate participle, the lack of standard complete sentuation lexicon, and the negation of words bring difficulty in implicit emotional recognition. The text data also contain rich and fine-grained information and thus become a difficult research point in natural language processing. This study proposes a hierarchical disconnected recurrent neural network to overcome the problem of lack of emotional information in implicit sentiment sentence recognition. The network encodes the words and characters in the sentence by using the disconnected recurrent neural network and fuses the context information of the implicit sentiment sentence through the hierarchical structure. By using the context information, the capsule network is used to construct different fine-grained context information for extracting high-level feature information and provide additional semantic information for emotion recognition. This way improves the accuracy of implicit emotion recognition. Experimental results prove that the model is better than some current mainstream models. The F1 value reaches 81.5%, which is 2 to 3 percentage points higher than those of the current mainstream models.
  Keywords: hierarchical interrupted circulation network; implicit emotion; capsule network; sentiment orientation identification.
  
  
• Greedy algorithm for image quality optimisation based on turtle-shell steganography  
  by Guo-Hua Qiu, Chin-Feng Lee, Chin-Chen Chang 
  Abstract: Information hiding, also known as data hiding, is an emerging field that combines multiple theories and technologies. In recent years, Chang and Liu et al. have proposed new data-hiding schemes based on Sudoku, a turtle-shell, etc. These proposed schemes have their own advantages in terms of visual quality and embedded capacity. However, the reference matrices used in these schemes are not optimal. Based on the characteristics of these schemes, Jin et al. (2017) employed particle swarm optimisation to select the reference matrix and achieved approximately optimal results in reducing the distortion of the stego-image. However, the complexity is high. In this paper, a turtle-shell matrix optimisation scheme is proposed using a greedy algorithm. The experimental results show that our proposed greedy algorithm is better than the particle swarm optimisation scheme at finding a near-optimal matrix and achieving better stego-image quality, and it outperforms the particle swarm optimisation scheme in terms of computational amount and efficiency.
  Keywords: data hiding; turtle-shell steganography; particle swarm optimisation; greedy algorithm.
  
  
• Client-side ciphertext deduplication scheme with flexible access control  
  by Ying Xie, Guohua Tian, Haoran Yuan, Chong Jiang, Jianfeng Wang 
  Abstract: Data deduplication with fine-grained access control has been applied in practice to realise data sharing and reduce the storage space. However, many existing schemes can only achieve server-side deduplication, which greatly wastes the network bandwidth even when the data transmitted is particularly large. Moreover, few existing schemes consider attribute revocation, in which the forward and backward secrecy cannot be guaranteed. To address the above problems, in this paper, we introduce a client-side ciphertext deduplication scheme with more flexible access control. Specifically, we divide the data owner into different domains and distribute corresponding domain keys to them through the secure channel, achieving PoW verification in client-side deduplication. Besides, we realise attribute revocation through the proxy re-encryption technology, which cannot preset the maximum number of clients in system initialisation. Security and performance analysis shows that our scheme can achieve desired security requirements while realising the efficient client-side deduplication and attribute revocation.
  Keywords: client-side deduplication; flexible access control; attribute revocation; random tag.
  
  

Special Issue on: PDCAT 2016 Parallel and Distributed Algorithms and Applications

• Data grouping scheme for multi-request retrieval in MIMO wireless communication  
  by Ping He, Zheng Huo 
  Abstract: The multi-antenna data retrieval problem refers to findng an access pattern (to retrieve multiple requests by using multiple antennae, where each request has multiple data items) such that the access latency of some requests retrieved by each antenna is minimised and the total access latency of all requests retrieved by all antennae keeps balance. So it is very important that these requests are divided into multiple groups for achieving the retrieval by using each antenna in MIMO wireless communication, called the data grouping problem. There are few studies focused on data grouping schemes applied to the data retrieval problem when the clients equipped with multi-antenna send multiple requests. Therefore, this paper proposes two data grouping algorithms (HOG and HEG) that are applied in data retrieval such that the requests can be reasonably classified into multiple groups. Through experiments, the proposed schemes have currently better efficiency compared with some existing schemes.
  Keywords: mobile computing; data broadcast; indexing; data scheduling; data retrieval; data grouping.
  
  
• Improved user-based collaborative filtering algorithm with topic model and time tag  
  by Liu Na, Lu Ying, Tang Xiao-jun, Li Ming-xia, Chunli Wang 
  Abstract: Collaborative filtering algorithms make use of interactions rates between users and items for generating recommendations. Similarity among users is calculated based on rating mostly, without considering explicit properties of users involved. Considering the number of tags of users can direct response the user preference to some extent, we propose a collaborative filtering algorithm using the topic model called UITLDA in this paper. UITLDA model consists of two parts. The first part is active user with its item. The second part is active user with its tag. We form the topic model from these two parts. The two topics constrain and integrate into a new topic distribution. This model not only increases the user's similarity, but also reduces the density of the matrix. In prediction computation, we also introduce time delay function to increase the precision. The experiments showed that the proposed algorithm achieved better performance compared with baseline on MovieLens datasets.
  Keywords: collaborative filtering; LDA; topic model; time tag.
  
  
• Improving runtime performance and energy consumption through balanced data locality with NUMA-BTLP and NUMA-BTDM static algorithms for thread classification and thread type-aware mapping  
  by Iulia Știrb 
  Abstract: Extending compilers such as LLVM with NUMA-aware optimisations significantly improves runtime performance and energy consumption on NUMA systems. The paper presents the NUMA-BTDM algorithm, which is a compile-time thread-type dependent mapping algorithm that performs the mapping uniformly, based on the type of each thread given by NUMA-BTLP algorithm following a static analysis on the code. First, the compiler inserts in the program code architecture-dependent code that detects at runtime the characteristics of the underlying architecture for Intel processors, and then the mapping is performed at runtime (using specific functions calls from the PThreads library) depending on these characteristics following a compile-time mapping analysis which gives the CPU affinity of each thread. NUMA-BTDM allows the application to customise, control and optimise the thread mapping and achieves balanced data locality on NUMA systems for C parallel code that combine PThreads-based task parallelism with OpenMP-based loop parallelism.
  Keywords: thread mapping; task parallelism; loop parallelism; compiler optimizations; NUMA systems; performance improvements; energy consumption improvements.
  
  
• Accumulative energy-based seam carving for image resizing  
  by Yuqing Lin, Jiawen Lin, Yuzhen Niu, Haifeng Zhang 
  Abstract: With the diversified development of the digital devices, such as computer, mobile phone and television, how to resize an image or video to adapt to different display screens has been a heated topic. Seam carving does well in image resizing at most times, however it sometimes produces discontinuity in the image content or impaired salient objects. Therefore, we propose an accumulative energy-based seam carving method for image resizing. We distribute the energy of each pixel on the seam to its adjacent eight-connected pixels to avoid the extreme concentration of seams. In addition, we add the image saliency and the edge information into the energy function to reduce the distortion. To compute more efficiently, we use parallel computing environment as well. Experimental results show that compared with the existing methods, our method can both avoid the discontinuity of image content and distortions as well as better maintain the shape of the salient objects.
  Keywords: image resizing; seam carving; optimal seam; accumulative energy; saliency detection; edge detection.
  DOI: 10.1504/IJCSE.2018.10014036
   
  

Special Issue on: Cyberspace Security Protection, Risk Assessment and Management

• Intrusion detection approach for cloud computing based on improved fuzzy c-means clustering algorithm  
  by Xuchong Liu, Jiuchuan Lin, Xin Su, Yi Zheng 
  Abstract: Recently, cloud computing has become more and more important on the internet. Meanwhile, network attackers aim at this platform, and launch various of attacks to threaten the security of cloud computing. Some researchers have proposed fuzzy C-means clustering algorithm (FCM) to detect such attack. However, FCM contains some limitations, such as low detection accuracy, low precision, and slow convergence speed when detecting intrusions under the cloud computing scenario. In this paper, we propose an intrusion detection approach based on an objective function optimisation FCM algorithm. This approach uses kernel function to improve optimisation ability of FCM algorithm. Then, the proposed approach uses Lagrange multiplier approach to calculate cluster centre and membership matrix, which is able to optimise the objective function of the FCM algorithm and reduce algorithm complexity. The simulation experiment shows that our approach can achieve higher detection accuracy and precision in detecting intrusion into a cloud computing network, and has great advantages in performance of convergence.
  Keywords: cloud computing; intrusion detection; network attack; objective function optimization; Lagrange multiplier approach.
  
  

Special Issue on: Recent Advances in the Security and Privacy of Multimedia Big Data in the Social Internet of Things

• Smart embarked electrical network based on embedded system and monitoring camera  
  by Mohammed Amine Benmahdjoub, Abdelkader Mezouar, Larbi Boumediene, Youcef SAIDI 
  Abstract: To improve the quality of life and its comfort with more security, the world of transport is moving towards all-electric. This imposes an embarked electrical network type operation; this network is based on parallel alternators connecting, which requires more energy and needs synchronisation with identical phases between alternators. In addition, some conditions must be respected to avoid energy crises and increase the efficiency of the system. To ensure the stability and protection of this type of system, the control will be performed by a reliable controller with remote control and monitoring of all data in real time. In this paper, we realise a prototype of protection and monitoring of electrical equipment using a Raspberry Pi as an intermediate embedded system and an RPi camera. In addition, the communication between the electrical system and the web application will be done by Json file or by data stored in the database. For any change in the desired values, the electrical protection system sends a message and a musical warning to the website in real time. In addition, the monitoring will be generated by the FIFO memory for image processing and the servomotor to control the direction of the RPi camera.
  Keywords: embarked electrical network; phase shift detector; remote control; embedded system; ethernet network; semantic web.
  
  
• Cognitive fog for health: a distributed solution for smart city  
  by Shu Chen, Nanxi Chen, Jiayi Tang, Xu Wang 
  Abstract: The Internet of Things (IoT) connects
  Keywords: internet of things; fog computing; distributed computing; smart city.
  
  

Special Issue on: ICCIDS 2018 Computational Intelligence and Data Science

• Statistical tree-based feature vector for content-based image retrieval  
  by Sushila Aghav-Palwe, Dhirendra Mishra 
  Abstract: The efficiency of any content-based image retrieval system depends on the extracted feature vectors of individual images stored in the database. The generation of compact feature vectors with good discriminative power is a real challenge in the image retrieval system. This paper presents the experimentation carried out to generate compact feature vectors for a colour image retrieval system based on image content. It has two stages of operation. In the first stage, the energy compaction property of image transforms is used; in the second stage, the statistical tree approach is used for feature vector generation. The performance of image retrieval is tested using an image feature database as per various performance evaluation parameters, such as precision recall crossover point along with newly proposed conflicting string of images. With different colour spaces, image transforms and statistical measures, the proposed approach achieves a reduction in the feature vector size with better discriminative power.
  Keywords: statistical tree; image retrieval; image transform; feature extraction; low level features.
  DOI: 10.1504/IJCSE.2019.10022604
   
  
• A benchmarking framework using nonlinear manifold detection techniques for software defect prediction  
  by Soumi Ghosh, Ajay Rana, Vineet Kansal 
  Abstract: Prediction of software defects in time improves quality and helps in locating the defect-prone areas accurately. Although earlier considerable methods were applied, actually none of those measures was found to be fool-proof and accurate. Hence, a newer framework includes a nonlinear manifold detection model, and its algorithm originated for defect prediction using different techniques of nonlinear manifold detection along with 14 different machine learning techniques on eight defective software datasets. A critical analysis cum exhaustive comparative estimation revealed that the nonlinear manifold detection model has a more accurate and effective impact on defect prediction than feature selection techniques. The outcome of the experiment statistically tested by Friedman and post hoc analysis using the Nemenyi test, which validates that the hidden Markov model along with the nonlinear manifold detection model, outperforms and is significantly different compared with other machine learning techniques.
  Keywords: dimensionality reduction; feature selection; Friedman test; machine learning; Nemenyi test; nonlinear manifold detection; software defect prediction; post hoc analysis.