Search

The physical characteristics of the massive spectrum signals carrying the communication information and the statistical laws of these characteristics also potentially reflect the communication behavior of the communication individuals and the intelligence information related to the communication behavior. Intercepting and cracking signal content usually faces enormous difficulties and costs, and more often, we are not able to crack the encrypted signal content. However, by studying the physical features extracted from the spectrum monitoring signals and the statistical laws of these features, it is also possible to dig out the hidden relationships between communication individuals and even the communication network structure, so as to analyze the communication behaviors of the communication individuals. Based on the characteristics of carrier frequency, bandwidth, power, signal monitoring time and direction information of spectrum monitoring signals, this paper identifies each spectrum signal and studies the distribution characteristics and statistical laws of massive spectrum monitoring signals in the column coordinate system. Due to the clustering of the spectrum signals generated by the sources in the power, monitoring time and direction, and the correlation of the spectrum signals generated by the two parties in the communication process, based on the improved density clustering algorithm, this paper proposes a method for mining the communication relationship between communication individuals from the spectrum monitoring data, and guesses and constructs the communication network structure by matching the communication individual with the communication relationship. Finally, we analyze the communication network structure mined from the spectrum monitoring data.

Big data generated from social media and smart mobile devices has been regarded as a key to obtain insights into human behavior and been extensively utilized for launching marketing activities. A successful marketing activity requires attracting high social popularity to their contents, since higher popularity usually indicates stronger influence, more fame and higher revenue. In this paper, we focus on the question of how to improve popularity of videos sharing on websites like YouTube in mobile computing environment. Obviously, composing high quality titles and tags is beneficial for viewers to discover videos of their interests and increase their tendency to watch more videos. However, it is not an easy task for uploaders, which is especially true since the screen is tight for most mobile devices. To this end, this paper proposes a novel hybrid method based on multi-modal content analysis that recommends keywords for video uploaders to compose titles and tags of their videos and then to gain higher popularity. The method generates candidate keywords by integrating techniques of textual semantic analysis of original tags and recognition of video content. On one hand, taking the original keywords of a video as input, the method obtains most relevant words from WordNet and related video titles gathered from the three top video sharing sites (YouTube, Yahoo Video, Bing Video). On the other hand, through recognizing video content with deep learning technology, the method extracts the entity name of video content as candidate keywords. Finally, a TF-SIM algorithm is proposed to rank the candidate keywords and the most relevant keywords are recommended to uploaders for optimizing the titles and tags of their videos. The experimental results show that the proposed method can effectively improve the social popularity of the videos as well as extend the length of video viewing time per playback.

With the rapid development of urbanization, the living standard has been improved on a continued basis for urban residents and their activities have become increasingly frequent. Therefore, it is of massive significance to study the hot spots of urban residents' activities and enforce effective planning and decision-making for urban and traffic departments. In this paper, the data is preprocessed in the first place. Then, the passengers' pick-up points and travel track points are extracted, and the statistical analysis method is employed to analyze the travel length and travel time of urban residents. Finally, an improved FCM algorithm is proposed. The conventional Fuzzy c-means (FCM) clustering algorithm is classed as a local optimal algorithm, and the number of clustering is made uncertain. In view of the shortcomings as mentioned above, an improved (FCM) clustering algorithm is suggested in this paper, which adopts adaptive distance norm and adds its own norm induction matrix to each cluster in order to ensure global optimization. The partition coefficient (PC), classification entropy (CE) and (XB) index are introduced to assist in determining the optimal number of clustering. According to the statistical analysis of GPS track data, the morning peak and evening peak in a day is 7:00-10:00 and 17:00-20:00, respectively. Cluster analysis is carried out for each time period and the whole day using the model proposed in this paper. The results show that the number of hot spots in each time period and the whole day is 12, 10, 8, 13, 13, respectively. The hot spots are distributed in the business center, office and residential areas, which are consistent with the actual situation. It shows that the model proposed in this paper can effectively and accurately mine the hot spots of urban residents' activities. It plays an important role in urban planning and commercial layout.

In this paper, a novel Mobile-Coverage scheme is proposed to improve the coverage and lifetime of hybrid sensor networks, and thus the target region can be completely covered with minimum number of mobile sensor nodes and the energy consumption can be reduced in hybrid sensor networks at the same time. Our scheme is applied to a hybrid network comprising of both static and mobile sensor nodes, which are randomly deployed in a rectangle area. We derived the minimum number of sensor nodes as well as the mobile nodes required to cover the target region completely. Compared to the existing works, the minimum number of mobile sensor nodes required in this paper is significantly smaller, which implies that the movement distance and energy consumption of mobile nodes in our scheme is reduced. Moreover, we proved that the construction of our scheme based on the Virtual Force and Voronoi Graph (VFVG) is correct, and then derived the optimal trade-off between the coverage of the target region and energy consumption of mobile nodes. Our scheme can achieve the maximum coverage which is 98.7% by minimum number of mobile sensor nodes. Meanwhile, a connectivity rate of 76.84% can also be achieved by our connectivity evaluation method. Finally, a new topology construction algorithm-Topology Construction and Maintenance based on Trust (TCMT) is created to further increase the lifetime of the sensor network. We also evaluated the performance of our scheme in terms of coverage and lifetime, and carried out extensive experiments to compare the coverage and energy consumption in our scheme and several previous schemes. The experimental results showed that our scheme is more efficient than the previous schemes simulated.

This paper presents a novel framework that automates the creation of a trained-classifier integrated circuit from a dataset. The framework accepts a dataset in a comma-separated value format and performs several processing steps to create a trained model. After creating the model, the framework generates a tree-based machine learning classifier in two formats; Extensible Markup Language (XML) and Verilog. We use the XML representation to present the hierarchy of the generated tree and the Verilog code as a hardware description language representation of the trained model. Our framework uses the Verilog code as an input to a Field programmable Gate Array (FPGA) design validation flow. Then, we automate the Application Specific Integrated Circuit (ASIC) flow implementation and build a customized classifier integrated circuit. The novelty of the proposed framework lies in bridging the gap between machine learning model training and its hardware design when dealing with machine learning classifier’s implementation. Our framework addresses several challenges related to the design automation and implementation of customized machine learning classifier chips from raw dataset files. We discuss these challenges in detail in this paper and explain how researchers can use our proposed framework to build low-cost, high-performance classifier chips. Our framework operates at 100 MHZ and achieves 80.79% average 10-fold cross-validation accuracy across five different datasets.

Nowadays, crowd analysis is one of the most important concepts that needs be relied upon, it contributes to decision making and ensuring the safety and security of the crowd. There are a variety of interesting research problems within the scope of crowd analysis including crowd tracking, crowd behaviour recognition and crowd counting. Crowd counting based on images and videos has been studied in past years. Nonetheless, estimating and detecting the number of human heads remains a challenging task due to occlusions, resolution, and lighting changes. This paper provides an overview and performance comparison of crowd counting techniques using convolutional neural networks (CNN) based on density map estimation. In this paper, we present a comprehensive analysis and benchmarking of crowd counting based on the UCF-QNRF dataset that contains the largest number of crowd count images and head annotations available in the public domain. We also show the density maps generation and their empirical evaluation along with performance comparison.

The purpose of this paper is to establish a new simultaneous localization and mapping (SLAM) framework, which can make better use of the spatial structure of the environment, to improve the robustness and reliability of the system. To locate, a good map should be not only a representation of the real environment, but also a highly compressed and intelligently organized geography information of the environment. Furthermore, the map structure should be stable and robust to inevitable errors. Inspired by human positioning ability, we developed a non-uniform network mapping and localization algorithm (NU-SLAM) based on the complex network theory. Landmarks are classified and sparsely connected according to their localization ability. Therefore, data association and information transmission can be carried out hierarchically with low computational cost and robustness to data association errors. In this paper, a new visual to the mapping and localization problem is proposed. Simulation results show that the algorithm can achieve good results.

Diabetic foot ulcers (DFUs) are a serious complication for people with diabetes. They result in increased morbidity and pressures on health system resources. Developments in machine learning (ML) offer an opportunity for improved care of individuals at risk of DFUs, to identify and synthesise evidence about the current uses and accuracy of ML in the interventional care and management of DFUs, and, to provide a reference for areas of future research. PubMed, Google Scholar, Web of Science and Scopus were searched using the Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies (PRISMA-DTA) guidelines for papers involving ML and DFUs. In order to be included, studies needed to mention ML, DFUs, and report relevant outcome measures regarding ML algorithm accuracy. Bias in included studies was assessed using the quality assessment tool for diagnostic accuracy (QUADAS-2). 37 out of 3769 papers were included after applying eligibility criteria. Included papers reported accuracy measures for multiple types of ML algorithms in DFU studies. Whilst varying across the ML algorithm used, all studies reported at least 90% accuracy compared to gold standards using a minimum of one reported ML algorithm for processing or recording data. Applications where ML had positive effects on DFU data analysis and outcomes include image segmentation and classification, raw data analysis and risk assessment. ML offers an effective and accurate solution to guide analysis and procurement of data from interventions which are designed for the care of DFUs in small samples and study conditions. Current research is limited, and, for the development of more applicable ML algorithms, future research should address the following: direct comparison of ML applications with current standards of care, health economic analyses and large scale data collection. There is currently no evidence to confidently suggest that ML methods in DFU diagnosis are ready for implementation and use in healthcare settings.

A method for estimating interest levels from behavior features via tensor completion including adaptive similar user selection is presented in this paper. The proposed method focuses on a tensor that is suitable for data containing multiple contexts and constructs a third-order tensor in which three modes are “products”, “users” and “user behaviors and interest levels” for these products. By complementing this tensor, unknown interest level estimation of a product for a target user becomes feasible. For further improving the estimation performance, the proposed method adaptively selects similar users for the target user by focusing on converged estimation errors between estimated interest levels and known interest levels in the tensor completion. Furthermore, the proposed method can adaptively estimate the unknown interest from the similar users. This is the main contribution of this paper. Therefore, the influence of users having different interests is reduced, and accurate interest level estimation can be realized. In order to verify the effectiveness of the proposed method, we show experimental results obtained by estimating interest levels of users holding books.

Social engineering has posed a serious security threat to infrastructure, user, data and operations of cyberspace. Nevertheless, there are many conceptual deficiencies (such as inconsistent conceptual intensions, a vague conceptual boundary, confusing instances, overgeneralization and abuse) of the term making serious negative impacts on the understanding, analysis and defense of social engineering attacks. In this paper, an in-depth literature survey is conducted, the original meaning of social engineering in cybersecurity is traced, the conceptual evolution and technical development are analysed systematically, and the conceptual problems are discussed. Based on above work, this paper attempts to address these conceptual deficiencies by proposing a more compatible and precise definition of social engineering in cybersecurity (SEiCS). This definition eliminates the conceptual inconsistencies, covers the mainstream conceptual connotations, clarifies the conceptual boundary, mitigates the overgeneralization and abuse, etc. Five analysis tables (i.e., the comparative analysis of the SEiCS definition vs. mainstream conceptual intensions in the conceptual evolution, the comparative analysis of the SEiCS definition vs. typical definitions in the literature, the analysis of confusing “social engineering cases”, the analysis of popular social engineering attack scenarios, and the analysis of social-engineering-based attacks) are provided to illustrate the performance of the proposed definition.

Item recommendation has become a significant means to help people discover interesting items. Meanwhile, plenty of reviews and ratings in recommender system can be utilized to relieve data sparsity problem. However, existing review-based approaches ignore the influence of static preference of user and the static characteristics of item, which could reflect long-term and stationary property, and guide feature extraction from reviews. Moreover, adaptive property, i.e., the importance of the historical records to each user and item, is not fully exploited in previous works. In this paper, we propose an Attention- based Adaptive Memory Network (AAMN) model to leverage historical reviews and ratings systemically. Specifically, we propose an attention mechanism guided by the static features to learn the importance of different historical records, for modeling the adaptive features of users and items. Notably, this paper is the first to bring static features into adaptively extracting semantic information from reviews, which can not only characterize user and item from a global view, but also assist to distinguish the importance of different reviews. In addition to the attention mechanism, we propose a non-linear feature fusing layer and a deep interaction layer to combine the static features and adaptive features, which capture underlying interactions among these features. To further improve prediction accuracy and training efficiency, we propose a dynamic sampling strategy for model training. We conduct extensive experiments on 16 benchmark datasets from Amazon and Yelp. The results demonstrate that our model outperforms the state-of-the-art models.

This paper focus on the problem of guaranteed cost control for a class of genetic regulatory networks with multiple time-varying discrete delays and multiple constant distributed delays. Firstly, a novel method is proposed to establish a sufficient condition for the existence of guaranteed cost controller. The sufficient condition includes only several simple inequalities, which can easily solved by standard tool softwares, such as MATLAB. Secondly, the desired guaranteed cost controller is designed based on the solution of these inequalities. Thirdly, the proposed method is also available to the stabilization problem of genetic regulatory networks under consideration. Finally, the results of two numerical examples demonstrate the applicability of theoretical results. Compared with the existing results, this present paper has three merits: (i) Do not require to construct any Lyapunov-Krasovskii functional; (ii) the class of genetic regulatory networks under consideration is more general; (iii) the designed controller can be easily realized.

The past couple of years have marked continued growth in the applications and services of the Internet of Things (IoT). This has attracted the attention of new operators as well as institutional, corporate, and private investors in every sector of the economy, and as a result, new businesses are springing up rapidly. These include many start-up companies that are producing various kinds of useful IoT devices and Smart Applications (smart apps). While this can be seen as a boost for innovation in the IoT, some of these companies produce IoT devices and smart apps with security vulnerabilities. In this paper, we propose the IoT Hardware Platform Security Advisor (IoT-HarPSecA), a security framework intended to provide support to such IoT producers. IoT-HarPSecA offers three functionality features, namely security requirement elicitation, security best practice guidelines for secure development, and above all, a feature that recommends specific LightWeight Cryptographic Algorithms (LWCAs) for both software and hardware implementations. Accordingly, IoT-HarPSecA is composed of three main components, namely Security Requirements Elicitation (SRE) component, Security Best Practice Guidelines (SBPG) component, and LightWeight Cryptographic Algorithms Recommendation (LWCAR) component, each of them servicing one of the aforementioned features. We implement a command-line tool in C++ to serve as an interface between users and the proposed framework. IoT-HarPSecA can be employed during the early stages of IoT systems design, and it can also be used to facilitate the implementation of security in existing IoT systems. This paper presents a detailed description, design, and implementation of the SRE, SBPG, and LWCAR components of the proposed framework. Using real-world practical scenarios, we show how IoT-HarPSecA can be used to elicit security requirements and recommend appropriate LWCAs based on user inputs. While a full performance evaluation of the SRE and SBPG components is beyond the scope of this paper, we present a detailed performance evaluation of the LWCAR component, which shows that IoT-HarPSecA can serve as a roadmap for secure IoT development.

This paper explores the immunological model and implements it in the domain of intrusion detection on computer networks. The main objective of the paper is to monitor, log the network traffic and apply detection algorithms for detecting intrusions within the network. The proposed model mimics the natural Immune System (IS) by considering both of its layers, innate immune system and adaptive immune system respectively. The current work proposes Statistical Modeling based Anomaly Detection (SMAD) as the first layer of Intrusion Detection System (IDS). It works as the Innate Immune System (IIS) interface and captures the initial traffic of a network to find out the first-hand vulnerability. The second layer, Adaptive Immune-based Anomaly Detection (AIAD) has been considered for determining the features of the suspicious network packets for detection of anomaly. It imitates the adaptive immune system by taking into consideration the activation of the T-cells and the B-cells. It captures relevant features from header and payload portions for effective detection of intrusion. Experiments have been conducted on both the real-time network traffic and the standard datasets KDD99 and UNSW-NB15 for intrusion detection. The SMAD model yields as high as 96.04% true positive rate and around 97% true positive rate using real-time traffic and standard data sets. Highly suspicious traffic detected in the SMAD model is further tested for vulnerability in the AIAD model. Results show significant true positive rate, closer to almost 99% of accurately detecting the file-based and user-based anomalies for both the real-time traffic and standard data sets.

Many power distribution methods of microgrid (MG) are all limited by fixed structure of MG and mathematical model, which will be out of effect once the structure changes. In order to solve the above problems, this paper proposed an adaptive hierarchical control method considering generation cost. Firstly, the distributed generators (DGs) of MG are considered as multi-agent system, and the secondary controller based on finite-time theory is constructed. Secondly, the improved gray wolf optimization (IGWO) algorithm is used as the tertiary controller to dynamically optimize the rated power of DG, and the objective function and constraints are established. Thirdly, the adaptive virtual impedance is introduced to realize accurate power sharing and the closed-loop control is formed by combining the current loop and voltage loop controllers. Finally, by dynamically optimizing the rated power, the on-line real-time optimal power distribution of MG is achieved, and the stability of the system is proved by the theories of multi-agent consistency and finite-time stability. The control method proposed in this paper accelerates the convergence speed of MG, and the calculation results have high accuracy. At the same time, the flexibility and reliability of MG are improved. The simulation model is established in Matlab/Simulink environment, and the simulation results show that the method is effective.

Agricultural diseases and insect pests are one of the most important factors that seriously threaten agricultural production. Early detection and identification of pests can effectively reduce the economic losses caused by pests. In this paper, convolution neural network is used to automatically identify crop diseases. The data set comes from the public data set of the AI Challenger Competition in 2018, with 27 disease images of 10 crops. In this paper, the Inception-ResNet-v2 model is used for training. The cross-layer direct edge and multi-layer convolution in the residual network unit to the model. After the combined convolution operation is completed, it is activated by the connection into the ReLu function. The experimental results show that the overall recognition accuracy is 86.1% in this model, which verifies the effectiveness. After the training of this model, we designed and implemented the Wechat applet of crop diseases and insect pests recognition. Then we carried out the actual test. The results show that the system can accurately identify crop diseases, and give the corresponding guidance.

With the rapid development of computer networks and multimedia technologies, images, which are important carriers of information dissemination, have made human cognition of things easier. Image recognition is a basic research task in computer vision, multimedia search, image understanding and other fields. This paper proposes a hierarchical feature learning structure that is completely automatically based on the original pixels of the image, and uses the K-SVD (K-Singular Value Decomposition) algorithm with label consistency constraints to train the discriminant dictionary. For different types of image data sets, the algorithm only extracts image blocks. After dense sampling, an efficient OMP (Orthogonal Matching Pursuit) encoder is used to obtain a layered sparse representation. The improved SIFT (Scale Invariant Feature Transform) algorithm is used to solve the difficult problem of multimedia visual image stereo matching. The feature point extraction and stereo matching of multimedia visual images, different scales and different viewpoint images are analyzed separately. Aiming at a large number of low-dimensional geometric features of 3D images, this paper studies the extraction and sorting strategies of low-dimensional geometric features of 3D images. A sparse representation method for 3D images is proposed, and the sparseness of image features is evaluated. This further improves the accuracy of 3D image representation and the robustness of 3D image recognition algorithms.

The public key encryption with equality test (PKEET) allows the cloud server to judge whether two different ciphertexts are generated by the same message without decryption. Through this technique, PKEET provides an effective solution for building secure outsourced databases, and has made some rich achievements. This paper combines multivariate public key encryption and equality test, and proposes the first multivariate public key encryption scheme with equality test (MPKEET), which inherits the advantages of both primitives. Moreover, the equality test algorithm proposed in this paper is based on a straight line. Compared with the schemes based on bilinear pairing, it is simpler and easier to implement. And our MPKEET scheme achieves desirable security, which can resist linearization equation attacks, differential attacks, XL attacks, Gröbner basis attacks and the attack of quantum computer, when appropriate parameters are selected.

With the increasing use of resource-constrained IoT devices, the number of IoT Botnets has exploded with many variations and ways of penetration. Nowadays, studies based on machine learning and deep learning have focused on dealing with IoT Botnet with many successes, and these studies have required relevant data during malware execution. For this, the sandbox environment and behavior collection tools play an essential role. However, the existing sandboxes do not provide adequate behavior data of IoT botnet such as the C&C server communication, shared libraries requirements. Moreover, these sandboxes do not support a wide range of CPU architectures, data is not exhaustively collected during executable file runtime. In this paper, we present a new practical sandbox, named V-Sandbox, for dynamic analysis of the IoT Botnet. This sandbox is an ideal environment for IoT Botnet samples that exhibit all of their malicious behavior. It supports the C&C servers connection, shared libraries for dynamic files, and a wide range of CPU architectures. Experimental results on the 6141 IoT Botnet samples in our dataset have demonstrated the effectiveness of the proposed sandbox, compared to existing ones. The contribution of this paper is specific to the development of a usable, efficient sandbox for dynamic analysis of resource-constrained IoT devices.

Street light are among the most common infrastructure in cities. Street lights and sensors can be combined to generate an interface of data collection. The analysis of massive data serves as an integral element of a smart city. This paper proposes a highly efficient system for the configuration, deployment, and management of smart street lights. The features of fast deployment and high scalability of the container-based system management result in virtual deployment. Additionally, for database design, NoSQL and in-memory databases are integrated to realize flexible data management. In terms of data transmission, this paper designs an asymmetric key and an SSH encrypted tunnel. Moreover, when all the services are connected, it conducts legitimacy validation via a token. Therefore, this system can help meet the demands for data throughput, low-latency, configuration, and realization of a smart city. It boasts high efficiency and security. Besides, it offers a flexible data storage and management service to facilitate the massive data processing of a smart city. With respect to experiments, this paper designs a street lighting simulation system with edge computing devices (consisting of a micro-controller, a sensor, and an IP camera) and a street lighting function. The system collects real-time sensed environmental data, enables live streaming of images, and offers an API for historical data query. This paper utilizes container-based virtualization to deploy all edge computing devices on the server and validates the feasibility of simultaneous operation of multiple container-based services on edge computing devices. This system has high commercial value.

The Byzantine general problem is the core problem that consensus algorithms are trying to solve, which is at the heart of the design of blockchains. As a result, we have seen numerous proposals of consensus algorithms in recent years, trying to improve the level of decentralization, performance, and security of blockchains. In our opinion, there are two most challenging issues when we consider the design of such algorithms in the context of powering blockchains in practice. First, the outcome of a consensus algorithm usually depends on the underlying incentive model, so each participant should have an equal probability of receiving rewards for its work. Secondly, the protocol should be able to resist network failures, such as cloud services shutdown, while maintaining high performance otherwise. We address these two critical issues in this paper. First, we propose a new metric, called fair validity, for measuring the performance of Byzantine agreements. Intuitively, fair validity provides a lower bound for the probability of acceptances of honest nodes' proposals. This is a strong notion of fairness, and we argue that it is crucial for the success of a blockchain in practice. We then show that any Byzantine agreement could not achieve fair validity in an asynchronous network, so we will focus on synchronous protocols. This leads to our second contribution: we propose a fair, responsive, and partition-resilient Byzantine agreement protocol able to tolerate up to 1/3 corruptions. As we will show in the paper, our protocol achieves fair validity and is responsive in the sense that the termination time only depends on actual network delay, as opposed to arbitrary, pre-determined time-bound. Furthermore, our proposal is partition-resilient. Last but not least, experimental results show that our Byzantine agreement protocol outperforms a wide variety of state-of-art synchronous protocols, combining the best from both theoretic and practical worlds.

Medical image synthesis receives much popularity in recent years, and ample medical images can be synthesized by diverse deep learning models to alleviate the problem of lack of data in many medical imaging utilizations. However, most medical image synthesis methods still incorporate the well-known pooling operation in their convolutional neural networks-based / generative adversarial networks-based models, from which image details will be inevitably lost due to the pooling operation. In order to tackle the above problem, improved capsule-based networks, in which no pooling operation is executed and spatial details of images can be effectively preserved thanks to the equivariance characteristics of capsule models, are proposed in this paper to synthesize arterial spin labeling images, for the first time. Technically, three important issues in constructing improved capsule-based networks, including the depth of basic convolutions, the layer of capsules, and the capacity of capsules, are thoroughly investigated. Comprehensive experiments made up of region-based / voxel-based partial volume corrections and dementia diseases diagnosis based on two different datasets are conducted. The superiority of improved capsule-based networks introduced in this paper is substantiated from the statistical point of view.

Convolutional Neural Network (CNN) models are a type of deep learning architecture introduced to achieve the correct classification of breast cancer. This paper has a two-fold purpose. The first aim is to investigate the various deep learning models in classifying breast cancer histopathology images. This study identified the most accurate models in terms of the binary, four, and eight classifications of breast cancer histopathology image databases. The different accuracy scores obtained for the deep learning models on the same database showed that other factors such as pre-processing, data augmentation, and transfer learning methods can impact the ability of the models to achieve higher accuracy. The second purpose of our manuscript is to investigate the latest models that have no or limited examination done in previous studies. The models like ResNeXt, Dual Path Net, SENet, and NASNet had been identified with the most cutting-edge results for the ImageNet database. These models were examined for the binary, and eight classifications on BreakHis, a breast cancer histopathology image database. Furthermore, the BACH database was used to investigate these models for four classifications. Then, these models were compared with the previous studies to find and propose the most state-of-the-art models for each classification. Since the Inception-ResNet-V2 architecture achieved the best results for binary and eight classifications, we have examined this model in our study as well to provide a better comparison result. In short, this paper provides an extensive evaluation and discussion about the experimental settings for each study that had been conducted on the breast cancer histopathology images.

Automatic text classification is a research focus and core technology in information retrieval and natural language processing. Different from the traditional text classification methods (SVM, Bayesian, KNN), the class-center vector method is an important text classification method, which has the advantages of less calculation and high efficiency. However, the traditional class-center vector method for text classification has the disadvantages that the class vector is large and sparse, and its classification accuracy is not high because of the lack of semantic information. To overcome these problems, this paper proposes a novel class-center vector model for text classification using dependencies and a semantic dictionary. We respectively use WordNet English semantic dictionary and Tongyici Cilin Chinese semantic dictionary to cluster the English or Chinese feature words in the class-center vector and to significantly reduce the dimension of class-center vector, thereby realizing a new class-center vector for text classification using dependencies and a semantic dictionary. Experiments show that, compared with traditional text classification algorithms, the improved class-center vector method has lower time complexity and higher accuracy on the 20Newsgroups English corpus, Fudan and Sogou Chinese corpus. This paper is an improved version of our NLPCC2019 conference paper.

Internet of Things (IoT) is an integration of the Sensor, Embedded, Computing, and Communication technologies. The purpose of the IoT is to provide seamless services to anything, anytime at any place. IoT technologies play a crucial role everywhere, which brings the fourth revolution of disruptive technologies after the internet and Information and Communication Technology (ICT). The Research & Development community has predicted that the impact of IoT will be more than the internet and ICT on society, which improves the well-being of society and industries. Addressing the predominant system-level design aspects like energy efficiency, robustness, scalability, interoperability, and security issues result in the use of a potential IoT system. This paper presents the current state of art of the functional pillars of IoT and its emerging applications to motivate academicians and researches to develop real-time, energy-efficient, scalable, reliable, and secure IoT applications. This paper summarizes the architecture of IoT, with the contemporary status of IoT architectures. Highlights of the IoT system-level issues to develop more advanced real-time IoT applications have been discussed. Millions of devices exchange information using different communication standards, and interoperability between them is a significant issue. This paper provides the current status of the communication standards and application layer protocols used in IoT with the detailed analysis. The computing paradigms like Cloud, Cloudlet, Fog, and Edge computing facilitate IoT with various services like data offloading, resource and device management, etc. In this paper, an exhaustive analysis of Edge Computing in IoT with different edge computing architectures and existing status are deliberated. The widespread adoption of IoT in society has resulted in privacy and security issues. This paper emphasizes on analyzing the security challenges, privacy and security threats, conventional mitigation techniques, and further scope for IoT security. The features like fewer memory footprints, scheduling, real-time task execution, fewer interrupt, and thread switching latency of Real-Time Operating Systems (RTOS) enables the development of time critical IoT applications. Also, this review offers the analysis of the RTOS's suitable for IoT with the current status and networking stack. Finally, open research issues in IoT system development are discussed.

Models are first-class elements in Model-Driven Engineering (MDE). In this paradigm, the most widespread approaches adopted by the development community are Object-Oriented and ontological, formalized using Unified Modeling Language (UML) and Resource Description Framework (RDF), respectively. However, Object Management Group (OMG) does not provide a specific standard language for validating UML models against Object Constraints Language (OCL) constraints; meanwhile, World Wide Web Consortium (W3C) has defined Shapes Constraint Language (SHACL) as a standard validation language. Although the transformation between UML and RDF can be performed at the structural level, no effort has been made to transform OCL to SHACL. This paper addresses the transformation of OCL and text-based constraints to SHACL shapes in the context of Common Grid Model Exchange Standard (CGMES), a UML-based standard for electric utilities in Europe. This paper presents several contributions to the software engineering community. First, solving the validation problem in a standardized way. Second, facilitating European Network of Transmission System Operators for Electricity (ENTSO-E) the construction of an ontology associated with the CGMES standard. Third, allowing developers to integrate the two complementary approaches. Finally, Promoting the adoption and integration of the ontological approach in the software community.

Entity disambiguation refers to the accurate inference of the real mention of an entity with the same name according to the context. Most existing studies focused on long texts, for short texts, the performance has been unsatisfactory due to sparsity. In this paper, we treat the entity disambiguation task as a classification problem. we propose a novel neural network-based capsule network and convolutional neural network for entity disambiguation, leveraging full semantic information of short text data. In particular, a self-attention mechanism is utilized to further filter the semantic information extracted from the capsule network. On the other hand, a convolutional neural network with combined pooling is established to capture semantics from another channel. In the end, the semantic features obtained by the above models are combined through a fully connected layer to complete the task of entity disambiguation. The experimental results on the CCKS 2019 entity linking dataset showed that the dual-channel hybrid network proposed in this paper achieved an F1-score of 88.04%, which is superior to that of the existing mainstream deep learning model, thereby verifying the effectiveness of the model.

Recently, many anomaly intrusion detection algorithms have been developed and applied in network security. These algorithms achieve high detection rate on many classical datasets. However, most of them failed to address two challenges: 1) imbalanced traffic data with limited network attack, 2) multiple data sources that are distributed in different terminals. In detail, those algorithms assume that there are sufficient network traffic data to train their models for intrusion detection. Due to the network attack traffic is always scarce in the real-world network, this assumption is difficult to satisfy in most cases. In this paper, we use Multi-Task Learning (MTL) and oversampling methods to address those challenges of network intrusion detection. Firstly, we use the MTL method to treat each terminal as a single task, and then use relevant information between different terminals to help learn every single task. Meanwhile, we use the oversampling method to overcome the minority problem of attacks. Through a series of experiments on the latest UNSW-NB15 and CICIDS2018 datasets, this paper verifies the effectiveness of MTL and oversampling methods for network intrusion detection with limited network attack data, where they achieve more than 90% detection rate in different experimental settings.

Spatial Crowdsourcing (SC) is a new valuable paradigm, relies on crowd workers to perform a set of spatial-temporal tasks at specific locations. It has garnered attention in collecting and processing social, environmental, and other spatio-temporal data by the contribution of individuals, communities and groups of workers in the physical world. The objective of SC is to outsource a set of spatio-temporal tasks to a set of workers, which requires the workers to be physically traveling to the tasks' locations in order to perform them, i.e., taking photos or collecting real time weather information at pre-specified location. Existing solutions require crowd workers to disclose their precise locations to untrustworthy service providers. Location updates and tracking in spatial crowdsourcing raise several privacy concerns in that malicious parties could snoop on crowd workers' whereabouts. Thus, the crowd workers' privacy could be compromised by disclosing their locations to untrusted and possibly malicious parties. This paper provides a novel framework called Dummies' Centroid (DCentroid), which aims at preserving location privacy for crowd workers in SC. The framework adapts an anonymous communication technique using a dummy based approach to generate dummy locations, i.e. decoy locations, and send their centroid points (pseudolocations) to service providers for processing. This paper theoretically analyzes the DCentroid framework and guarantees the crowd workers' privacy, while preserving the functionality of SC, such as the success rate of task assignments, worker travel distances, and system overhead. Practical experimentation on real-world datasets shows that the DCentroid framework protects the crowd workers' location privacy without affecting the various performance parameters of task assignment.

In the current banking systems and business processes, the permission granted to employees is controlled and managed by the configured access control methods, in which static role-based models focus on access to information and functions. The deployed configuration is not reviewed/updated systematically and is handled manually by managers. Consequently, banks and companies are looking for systems and applications to automate and optimize their business processes and data management intelligently. In this context, the notion of integrating machine learning (ML) techniques in banking business processes has emerged. In order to build an intelligent and systematic solution, we combine in this paper ML and dynamic authorization techniques to enable performance-based policy evaluation into the banking teller process, where policies adapt to the changes recognized by the ML model. The objective of this work is to focus on the banking teller process that may be generalized to other operational banking processes. In this context, we propose in this paper a new model providing Intelligent Performance-Aware Adaptation of Roles and Policy Control using a support vector machine (SVM). We demonstrate that our model is capable of assessing the deployed control policies and updating them systematically with new roles and authorization levels based on tellers' performance, work history, and system constraints. We evaluated different machine learning models on a real dataset generated from a real-life banking environment. Experimental results explore the relevance and efficiency of our proposed scheme in terms of prediction accuracy, required authorizations, transaction time, and employees' working hours.

Communication-assisted protection and self-healing control (CAPSC) is perceived as a new alternative to ensure reliable operation of distribution networks (DNs) because of its prominent advantages of information sharing. Although the application of CAPSC based on IEC 61850 has been verified in substations, it has not yet been effectively promoted in DNs. To fully exert its advantages in DNs, a novel scheme of IEC 61850-based CAPSC is proposed in this paper. The proposed scheme enable samplings to be synchronized by propagation delay compensation without external clocks. Adaptive execution algorithms of protection and self-healing control are introduced considering changeable topology and configuration conditions. The coordination between protection and self-healing control in this scheme is simplified by component state descriptions instead of event signals. The scheme proposed in this paper has been incorporated in an engineering field application. A closed-loop test has been conducted on a real-time digital simulator (RTDS) according to the field connections and parameters to verify the rationality and effectiveness of the proposed scheme.

EA (Enterprise Architecture) visualization methodologies have been explored by researchers and engineers to conduct EA modeling. The objectives of EA modeling are to clarify enterprise strategies, visualize business processes, and model information systems to manage resources, improve organization structure, adjust information strategy, and create new business value. Therefore, EA models can be broadly applied in various fields. For example, the applications include business modeling, information system architecture design, technology infrastructure configuration, software maintenance, and system security analysis. As the primary source of information, EA models are of paramount importance to researchers, architects, and developers. However, up to now, the purpose and means of these EA visualization methods have never been systematically analyzed and discussed, and a generalized EA visualization methodology with the ability to meet different demands is needed. The paper narrows this gap by conducting a systematic literature review on enterprise architecture visualization methodologies. In this study, 112 papers are retrieved by a manual search in 5 academic databases, a systematic literature review on EA visualization is explained to show a systematized category of visualization approaches, and then a general visualization approach is proposed by systematically reviewing the papers. Finally, the paper is concluded by discussing the contributions and limitations of the study.

Breast cancer is one of the most dangerous diseases and the second largest cause of female cancer death. Breast cancer starts when malignant, cancerous lumps start to grow from the breast cells. Self-tests and Periodic clinical checks help to early diagnosis and thereby improve the survival chances significantly. The breast cancer classification is a medical method that provides researchers and scientists with a great challenge. Neural networks have recently become a popular tool in cancer data classification. In this paper, Deep Learning assisted Efficient Adaboost Algorithm (DLA-EABA) for breast cancer detection has been mathematically proposed with advanced computational techniques. In addition to traditional computer vision approaches, tumor classification methods using transfers are being actively developed through the use of deep convolutional neural networks (CNNs). This study starts with examining the CNN-based transfer learning to characterize breast masses for different diagnostic, predictive tasks or prognostic or in several imaging modalities, such as Magnetic Resonance Imaging (MRI), Ultrasound (US), digital breast tomosynthesis and mammography. The deep learning framework contains several convolutional layers, LSTM, Max-pooling layers. The classification and error estimation that has been included in a fully connected layer and a softmax layer. This paper focuses on combining these machine learning approaches with the methods of selecting features and extracting them through evaluating their output using classification and segmentation techniques to find the most appropriate approach. The experimental results show that the high accuracy level of 97.2%, Sensitivity 98.3%, and Specificity 96.5% has been compared to other existing systems.

With the advancements in machine and deep learning algorithms, the envision of various critical real-life applications in computer vision becomes possible. One of the applications is facial sentiment analysis. Deep learning has made facial expression recognition the most trending research fields in computer vision area. Recently, deep learning-based FER models have suffered from various technological issues like under-fitting or over-fitting. It is due to either insufficient training and expression data. Motivated from the above facts, this paper presents a systematic and comprehensive survey on current state-of-art Artificial Intelligence techniques (datasets and algorithms) that provide a solution to the aforementioned issues. It also presents a taxonomy of existing facial sentiment analysis strategies in brief. Then, this paper reviews the existing novel machine and deep learning networks proposed by researchers that are specifically designed for facial expression recognition based on static images and present their merits and demerits and summarized their approach. Finally, this paper also presents the open issues and research challenges for the design of a robust facial expression recognition system.

The most burning topic of today, calls for a holistic solution that is reliable, secure, privacy preserved, cost effective Cloud storage that can tide over the turbulent conditions of the rapidly budding digital storage technologies. This send an outcry for a devoted solution, in the form of an individualized, patient-centric care - IoMT that augments precise disease identifications, decrease in errors, reduction in costs of care through the support of technology, allows patients to direct health information data to doctors,manage drugs, keep Personal Health Records, caters to remote medical supports Care, provides proactive approach to preserving Good Health, improves and Accelerates Clinician Workflows, empowers extreme connectivity due to better automation and perceptions in the DNA of IoMT functions. But IoMT adoption is like a rose with thorns like constraints of increased administrative costs, deficiency of universal data access, present-day electronic medical records. The BCT is used in the framework to overcome the security issues of IoMT through the use of latest encryptions. Furthermore, this framework harnesses the benefits of Block Chain like reduced cost, speed, automation, immutability, near-impossible loss of data, permanence, removal of intermediaries, decentralization of consensus, legitimate access to health data, data safekeeping, accrual-based imbursement mechanisms, and medical supply chain efficacy. The outcomes in this paper are (i)A systematic investigation of the current IoMT, Block Chain and Cloud Storage in Health Care;(ii) Explore the challenges and necessities for the confluence of Block Chain (BC), Internet of Medical Things (IoMT), Cloud Computing (CC);(iii)Formulate the requirements necessary for the real-time remote Health Care of one-to-one care structure, which, supports the vital functions that are critical to the Patient Centric Health Care;(iv) Design and develop a novel BC IoMT U6 HCS (Block Chain based Internet of Medical Things for Uninterrupted, Ubiquitous, User-friendly, Unflappable, Unblemished, Unlimited Health Care Services) Layered Architecture, to support the vital functions critical for Patient Centric Health Care and (v) Implement and test with the previous established and proven techniques. The integrity of the Layered Architecture is validated with the already existing ones in terms of audit performances. The results from the Layered Architecture are validated and are proven to be competent in achieving safe auditing and surpass the former ones. The technology is in the sprouting phases, it is perilous that affiliates of the Health Care community realize the rudimentary ideas behind Block Chain, and detect its feasible impact on the future of patient centric medical care. Finally, and most importantly, this paper also gives a panoramic view on the current research status, and imminent directions of Secure Internet of Medical Things Using Block Chain.

Solar Photovoltaic System (SPV) is one of the growing green energy sources having immense penetration in the national grid as well as the off-grid around the globe. Regardless of different solar insolation level at various regions of the world, SPV performance is also affected by several factors: conversion efficiency of PV cell technology, ambient temperature and humidity, soiling and seasonal/weather patterns. The rise in PV cell temperature and soiling is found to be detrimental issues regarding power plant performance and life expectancy leading alterations in the levelised cost of energy (LCoE). In this paper, authors present a short glance about factors affecting the performance of photovoltaic modules and re-discuss their usability in cleaning intervention decision-making models. With some highlights on the essence of cleaning to mitigate the soiling issues in PV power plants, this paper presents the existing cleaning techniques and practices along with their evaluations. The need for an optimal cleaning intervention by using advanced scientific tools rather than by visual inspection is drawing the attention of PV experts. The authors finally suggest a schematic of a decision-making model which involves the use of probable parameters, data processing techniques and machine learning tools. The implementation of data science and machine learning in a solar PV panel cleaning system could be a remarkable advancement in the field of renewable energy.

Text generation is a basic work of natural language processing, which plays an important role in dialogue system and intelligent translation. As a kind of deep learning framework, Generative Adversarial Networks (GAN) has been widely used in text generation. In combination with reinforcement learning, GAN uses the output of discriminator as reward signal of reinforcement learning to guide generator training, but the reward signal is a scalar and the guidance is weak. This paper proposes a text generation model named Feature-Guiding Generative Adversarial Networks (FGGAN). To solve the problem of insufficient feedback guidance from the discriminator network, FGGAN uses a feature guidance module to extract text features from the discriminator network, convert them into feature guidance vectors and feed them into the generator network for guidance. In addition, sampling is required to complete the sequence before feeding it into the discriminator to get feedback signal in text generation. However, the randomness and insufficiency of the sampling method lead to poor quality of generated text. This paper formulates text semantic rules to restrict the token of the next time step in the sequence generation process and remove semantically unreasonable tokens to improve the quality of generated text. Finally, text generation experiments are performed on different datasets and the results verify the effectiveness and superiority of FGGAN.

This paper investigates the adaptive distributed synchronization control problem of multi-agent systems, where the information transmission in the network is subject to unknown constant signal transmission delay. Firstly, an adaptive distributed control scheme is designed for a class of certain multi-agent systems to guarantee the synchronization among all agents is achieved. In the scheme, the unknown constant delay can be estimated online. Furthermore, compared with some existing works where some global information are required to design the controllers for the agents, the scheme proposed in our paper does not need the global information. By algebraic graph theory with Lyapunov stability theory, the whole closed-loop system stability analysis is provided. Then, the method is extended to a class of uncertain nonlinear MASs. Finally, genetic regulatory networks (GRNs) are used to show the effectiveness of the proposed design scheme.

With the development of the Cyber-Physical-Social Systems(CPSS), a large number of multidimensional time series have been generated in today's world, such as: sensor data for industrial equipment operation, vehicle driving data, and cloud server operation and maintenance data and so on. The key task of Cloud-Fog-Edge Computing in managing these systems is how to detect anomalous data in a specific time series to facilitate operator action to solve potential system problems. So multidimensional time series outlier detection become an important direction of CPSS data mining and Cloud-Fog-Edge Computing research, and it has a wide range of applications in industry, finance, medicine and other fields. This paper proposes a framework called Multidimensional time series Outlier detection based on a Time Convolutional Network AutoEncoder (MOTCN-AE), which can detect outliers in time series data, such as identifying equipment failures, dangerous driving behaviors of cars, etc. Specifically, this paper first uses a feature extraction method to transform the original time series into a feature-rich time series. Second, the proposed TCN-AE is used to reconstruct the feature-rich time series data, and the reconstruction error is used to calculate outlier scores. Finally, the MOTCN-AE framework is validated by multiple time series datasets to demonstrate its effectiveness in detecting time series outliers.

Passenger assignment of rail transit has recently attracted increasing research interest due to its potential applications in large-scale intelligent transportation systems. In the rail transit system, the foundation of passenger assignment is passengers' origin and destination demand (OD matrix). However, due to the nature of stochastic of the short-term dynamic OD matrix, how to accurately predict the distribution of passenger travel spatio-temporally is still an open challenge. In this paper, combined multisource data with deep learning method is proposed to improve prediction of dynamic OD matrix accuracy. Firstly, multisource data such as smart card data, weather data and mobile phone data are introduced. And after quantitative analysis of the influencing factors, choosing 31 features as model inputs. Secondly, considering the superiority of Long Short-term Memory Network in time series, we improve the structure of LSTM by redesigning the hidden layer and neuron, in view of the spatio-temporal characteristics of spatio-temporal Long Short-term Memory Network (STLSTM) of rail transit passenger flow. Finally, using the Beijing subway network which had 54,056OD for verification. Extensive experiments and evaluations on a large-scale dataset well demonstrate the superiority of STLSTM over commonly used prediction models and standard LSTM model for short-term prediction of dynamic OD matrix. In addition, the application method of multisource data in OD prediction in this paper can deal with more data from other sources to further improve the information exploit effect on passenger flow law.

One of the most amazing functions of the human auditory system is the ability to detect all kinds of sound events in the environment. With the technologies and hardware advances, polyphonic Sound Event Detection (SED) can be developed to mimic the ability of the human auditory system. However, the development of a SED system is no trivial task, and several different factors often hinder accuracy. Although there are several overview papers available, most of them only provide a theoretical overview of algorithms used with little discussion. Thus, to the best of the authors' knowledge, there is no comprehensive review that covers this particular domain. Therefore, this paper aims to provide an in-depth discussion of different methodologies proposed by various authors that include the features used, detection algorithms, and their corresponding accuracy and limitations. Additional information on possible trends is also discussed that can be useful for future development works.

Over the last few decades, sustainable computing has been widely used in areas like social computing, artificial intelligence-based agent systems, mobile computing, and Internet of Things (IoT). There are social, economic, and commercial impacts of IoT on human lives. However, IoT nodes are generally power-constrained with data transmission using an open channel, i.e., Internet which opens the gates for various types of attacks on them. In this context, several efforts are initiated to deal with the evolving security issues in IoT systems and make them self-sufficient to harvest energy for smooth functioning. Motivated by these facts, in this paper, we explore the evolving vulnerabilities in IoT devices. We provide a state-of-the-art survey that addresses multiple dimensions of the IoT realm. Moreover, we provide a general overview of IoT, Sustainable IoT, its architecture, and the Internet Engineering Task Force (IETF) protocol suite. Subsequently, we explore the open-source tools and datasets for the proliferation in research and growth of IoT. A detailed taxonomy of attacks associated with various vulnerabilities is also presented in the text. Then we have specifically focused on the IoT Vulnerability Assessment techniques followed by a case study on sustainability of Smart Agriculture. Finally, this paper outlines the emerging challenges related to IoT and its sustainability, and opening the doors for the beginners to start research in this promising area.

Single-stage object detectors are quick and highly accurate. Based on the way training model is developed, single-stage object detectors either adopt a training model based on a pre-trained backbone network model, or a model trained from the scratch. The pre-trained backbone network model is associated with the propagation sensitivity both in classification and detection. This leads to deviations in learning goals, and results in an architecture which is limited by the classification network, hence not easy to modify. Training from the scratch is not as efficient as using a pre-trained network, mainly due to the limitations of the predefined network system. In this paper, we combine these two approaches to overcome the above-mentioned shortcomings. In our proposed method a top-down concatenated feature pyramid is built upon a basic FSSD network. The experiments in this paper are conducted on MS COCO and PASCAL VOC data sets. Moreover, we apply VGG16 as the backbone network to further indicate the effectiveness of our proposed method which reaches 33.1 AP on MS COCO benchmark.

Context: JavaScript (JS) is an often-used programming language by millions of web pages and is also affected by thousands of malicious attacks. Objective: In this investigation, we provided a general view and a quick understanding of JavaScript Malware Detection (JSMD) research reported in the scientific literature from several perspectives. Method: We performed a Systematic Literature Review (SLR) and quality analysis of published research articles on the topic. We investigated 32 articles published between the year 2009 to the year 2019. Results: Selected 32 papers explained in this article reflect the outline of what was published so far. One of our key findings is the performance of Machine Learning (ML) based detection models were relatively higher than others. We also found that only a few papers were able to achieve high scores according to the quality assessment criteria. Conclusion: In this SLR, we summarized and synthesized the existing JSMD studies to identify the previous research practices and also to shed light on future guidelines in the malware detection space. This study will guide and help future researchers to investigate the previous literature efficiently and effectively.

Internet of Things (IoT) and smart computing technologies have revolutionized every sphere of 21st century humans. IoT technologies and the data driven services they offer were beyond imagination just a decade ago. Now, they surround us and influence a variety of domains such as automobile, smart home, healthcare, etc. In particular, the Agriculture and Farming industries have also embraced this technological intervention. Smart devices are widely used by a range of people from farmers to entrepreneurs. These technologies are used in a variety of ways, from finding real-time status of crops and soil moisture content to deploying drones to assist with tasks such as applying pesticide spray. However, the use of IoT and smart communication technologies introduce a vast exposure to cybersecurity threats and vulnerabilities in smart farming environments. Such cyber attacks have the potential to disrupt the economies of countries that are widely dependent on agriculture. In this paper, we present a holistic study on security and privacy in a smart farming ecosystem. The paper outlines a multi layered architecture relevant to the precision agriculture domain and discusses the security and privacy issues in this dynamic and distributed cyber physical environment. Further more, the paper elaborates on potential cyber attack scenarios and highlights open research challenges and future directions.

Congestion often hinders human mobility. This problem occurs due to the constant increase in vehicles every year. Reliable predictions of traffic conditions would allow drivers to choose their routes to avoid traffic jams while providing the police with traffic management strategies. Therefore, this paper tests the ability of various machine learning methods to predict traffic conditions. The study assesses Neural Networks, Bayes Classifier, Decision Trees, SVM, Deep Neural Network, and Deep Learning. Of these methods, the Decision Tree, Deep Neural Network, and Bayes Classifier show the highest performance in predicting traffic conditions using static data testing. However, in dynamic testing to assess the growth of knowledge, the performance of the Knowledge-Growing Decision Tree tends to decrease as the training data grows. Its performance decreased 3.89 points (88.24% to 84.35%) in accuracy, and 7.55 points (76.25% to 68.70%) for each precision, recall, and F1 Score. Conversely, the Knowledge-Growing Deep Neural Network and Bayes Classifier had a better performance than Decision Tree. The performances of Knowledge-Growing Deep Neural Network increased slightly by 0.35 points (93.38% to 93.73%) for accuracy and 0.69 points (86.77% to 87.64%) in other measurements. Although its performance increased, the processing time takes very long, namely 139452.76 seconds and 318832.80 seconds for sub-scheme (a) and (b), respectively. Meanwhile, the Knowledge-Growing Bayes Classifier offers a greater performance increase of 2.3 points (80.52% to 82.82%) for the accuracy and 4.6 points (65.63% to 61.03%) for the other performance measurements. In addition, it also scored better for processing time, as predictions only take 3 seconds using sub-scheme (a), and 7 seconds when using sub-scheme (b). Therefore, the paper proposes the Knowledge-Growing Bayes Classifier to predict rapidly changing traffic conditions. This method outperform the others. These can be attributed to its ability to 1) adjust to ever-changing the traffic conditions; 2) predict the result as soon as the data are acquired; and 3) make decentralized predictions.

All major banks in the USA and around the world offer remote check deposit services. Consumers can use their smart phones to deposit checks remotely. This new online check truncation system is vulnerable to a wide range of attacks, including digital check forgery. Shifting trust from a human teller or an automated teller machine (ATM) to a smart device (cell phone) provides new attack surfaces. This paper exploits security vulnerabilities in the existing remote check deposit system and presents an attack vector for existing remote check truncation systems. The proposed attack vector exploits vulnerabilities in the untrusted client-side check-deposit system that enables an attacker to instrument the check deposit application library. The instrumented library allows the attacker to induce digital check forgery with minimized tampering artifacts. It has been observed through this investigation that digital check forgery-based attacks are more powerful than conventional paper-based check forgery attacks. The effectiveness of these attacks is evaluated by targeting three leading banks in United States, finding that all three of the targeted banks are vulnerable to the proposed attacks. A set of countermeasures based on digital check verification is also proposed to combat digital check forgery attacks on existing remote check deposit systems. The proposed countermeasures rely on tamper detection in digital images and expert-system based decision fusion. The effectiveness of the proposed framework is evaluated using tampered check images. The tampered images used for performance evaluation also include set of tampered images used for successfully attacking the remote check deposit systems(being using by leading banks around the world today). Experimental results show that the proposed expert system-based framework is capable of detecting digital check forgery attacks.

With the popularity of online social networks, researches on dynamic node classification have received further attention. Dynamic node classification also helps the rapid popularization of online social networks. This paper proposes a particle competition model named DPP to complete the dynamic node classification. Existing node classification models based on particle competition do not perform well in terms of accuracy. Hence, we formulate a unique particle competition framework to make the node classification more effective. In addition, for applying the model in the dynamic network, based on the dynamic characteristics of the model, we have added an automatic update strategy of the source node to the model. The particles in the new model perform the steps of walking, splitting, and jumping according to the method introduced in this paper. Then, the domination matrix of the network has been changed with particle movements continuously. Although the particles randomly walk at the micro-level, the model can converge to obtain the node classification results. Finally, simulation results show both the effectiveness and superiority of our proposed node classification model with the comparison of other major particle competition models and dynamic node classification methods. Based on the above contributions, the proposed model may have compelling applications in the context of community detection and network embedding, etc.

Molecular communication is a novel communication paradigm, which has prospective potential applications in many fields. Obtaining knowledge of the diffusive molecular channel is important for the design of the molecular communication systems as well as proper working of many application systems. In this paper, we focus on a kind of molecular communication system with flow drift, which is modeled as an inverse Gaussian distributed channel. The maximum likelihood estimation method is applied to obtain the estimator of parameters such as medium velocity, propagation distance, and diffusion coefficient. This paper also derives the closed-form expressions of the Cramer-Rao lower bounds. The performances of our channel parameter estimators are validated in MATLAB and the results confirm its effectiveness.

This paper aims at visualizing deep convolutional neural network interpretations for aerial imagery and understanding how these interpretations change across datasets or when network weights are damaged. Our visualization results offer insights on the generalization power and resilience of commonly used networks, such as VGG16, ResNet50, and DenseNet121. Our experiments on the AID and the UCM aerial datasets demonstrate the emergence of object and texture detectors in convolutional networks commonly used for classification. We further analyze these interpretations when the network is trained on one dataset and tested on another to demonstrate the robustness of feature learning across aerial datasets. We also explore the shift in interpretations when performing transfer learning from an aerial dataset (AID) to a generic object dataset (MS-COCO). These results illustrate how transfer learning benefits the network’s internal representations. For analyzing the effects of damages on activation maps, our work proposes to simulate damages by randomly zeroing network weights at different levels of the network. The paper carries out experiments with retraining the network to check if it can recover the lost interpretations. Visualizing changes in the neural network’s interpretation when the undamaged weights are updated, allows us to assess the resilience of a network visually. Finally, we propose a new metric for the quantitative assessment of network resilience.