Search

The Internet of Things (IoT) devices, networks, and applications have become an integral part of modern societies. Despite their social, economic, and industrial benefits, these devices and networks are frequently targeted by cybercriminals. Hence, IoT applications and networks demand lightweight, fast and flexible security solutions to overcome these challenges. In this regard, Artificial Intelligence (AI)-based solutions with big data analytics can produce promising results in the field of cybersecurity. This article proposes a lightweight Dense Random Neural Network (DnRaNN) for intrusion detection in the IoT. The proposed scheme is well suited for implementation in resource-constrained IoT networks due to its inherent improved generalization capabilities and distributed nature. The suggested model was evaluated by conducting extensive experiments on a new generation IoT security dataset ToN_IoT. All the experiments were conducted under different hyperparameters and the efficiency of the proposed DnRaNN was evaluated through multiple performance metrics. The findings of the proposed study provide recommendations and insights in binary class and multiclass scenarios. The proposed DnRaNN model attained attack detection accuracy of 99.14% and 99.05% for binary class and multiclass classifications, respectively.

The Adaptive Object Model (AOM) is an architectural style in which domain entity types are represented as instances that can be changed at runtime. It can be used to achieve higher flexibility in domain classes. Due to AOM entities having a distinct structure, they are not compatible with most popular frameworks, especially those that use reflection and code annotations. To solve such limitations, this study aims to propose a model that enables the reuse of frameworks designed for classic object-oriented domain models in an AOM application. The proposed model uses dynamically-generated adapters for AOM entities that encapsulate them in a class with the format expected by the frameworks. A reference implementation was developed in the Esfinge AOM RoleMapper framework to evaluate the viability of the proposed model. Initially, to evaluate the solution feasibility, a case study was carried out using the Hibernate framework. Further, an experiment was conducted to assess how the participants perceived the framework functionality reuse through the proposed model. The feasibility study revealed that the solution could be applied in a complex setting for the chosen object-relational mapping frame. It raised some difficulties that can be addressed in future studies. In the experiment, the development time did not present a significant difference compared to the competing approach. Despite the considerable learning curve, most participants considered that the proposed approach has more advantages than the alternative. Based on the evaluations, we can conclude that the proposed model can be successfully employed to use AOM entities with frameworks that were not designed for AOM applications. The possibility of reusing existing frameworks can reduce the effort required to adopt an AOM architecture and, consequently, be a facilitator in implementing more flexible and adaptive approaches.

Neural Style Transfer (NST) is a class of software algorithms that allows us to transform scenes, change/edit the environment of a media with the help of a Neural Network. NST finds use in image and video editing software allowing image stylization based on a general model, unlike traditional methods. This made NST a trending topic in the entertainment industry as professional editors/media producers create media faster and offer the general public recreational use. In this paper, the current progress in Neural Style Transfer with all related aspects such as still images and videos is presented critically. The authors looked at the different architectures used and compared their advantages and limitations. Multiple literature reviews focus on either the Neural Style Transfer (of images) or cover Generative Adversarial Networks (GANs) that generate video. As per the authors’ knowledge, this is the only research article that looks at image and video style transfer, particularly mobile devices with high potential usage. This article also reviewed the challenges faced in applyingvideo neural style transfer in real-time on mobile devices and presents research gaps with future research directions. NST, a fascinating deep learning application, has considerable research and application potential in the coming years.

The exponential growth in the adoption of drones in various industrial applications has created a demand for professionals experienced in designing, programming, operating, and piloting drones. This paper reports on an active learning-based instructional approach that prepares students for careers in the drone industry. The approach focuses on three foundational pillars of drone careers: 1) drone programming using Python, 2) designing and fabricating drones using Computer-Aided Design (CAD) and rapid prototyping, and 3) the US Federal Aviation Administration (FAA) Part 107 Commercial small Unmanned Aerial Vehicles (sUAVs) pilot test. The approach was implemented and refined over four semesters in an undergraduate class attended by engineering and computer science students at a research institution located in the United States. A case study approach using multiple assessment methods was implemented to examine students’ gains in skills and knowledge and attitudes towards an active learning-based approach for drone education. Assessment methods included direct formative (FAA quizzes and programming assignments), direct summative (capstone project), indirect quantitative (survey of learning gains), and indirect qualitative (focus group interviews and capstone project process videos) tools. Assessment results noted in this paper indicate that students were actively engaged during the course and made positive gains in their learning outcomes.

Security and safety is a big concern for today’s modern world. For a country to be economically strong, it must ensure a safe and secure environment for investors and tourists. Having said that, Closed Circuit Television (CCTV) cameras are being used for surveillance and to monitor activities i.e. robberies but these cameras still require human supervision and intervention. We need a system that can automatically detect these illegal activities. Despite state-of-the-art deep learning algorithms, fast processing hardware, and advanced CCTV cameras, weapon detection in real-time is still a serious challenge. Observing angle differences, occlusions by the carrier of the firearm and persons around it further enhances the difficulty of the challenge. This work focuses on providing a secure place using CCTV footage as a source to detect harmful weapons by applying the state of the art open-source deep learning algorithms. We have implemented binary classification assuming pistol class as the reference class and relevant confusion objects inclusion concept is introduced to reduce false positives and false negatives. No standard dataset was available for real-time scenario so we made our own dataset by making weapon photos from our own camera, manually collected images from internet, extracted data from YouTube CCTV videos, through GitHub repositories, data by university of Granada and Internet Movies Firearms Database (IMFDB) imfdb.org. Two approaches are used i.e. sliding window/classification and region proposal/object detection. Some of the algorithms used are VGG16, Inception-V3, Inception-ResnetV2, SSDMobileNetV1, Faster-RCNN Inception-ResnetV2 (FRIRv2), YOLOv3, and YOLOv4. Precision and recall count the most rather than accuracy when object detection is performed so these entire algorithms were tested in terms of them. Yolov4 stands out best amongst all other algorithms and gave a F1-score of 91% along with a mean average precision of 91.73% higher than previously achieved.

Error checking is easy and inexpensive in the initial stages as compared to later stages due to when the development cycle precedes the development cost and efforts also increase. UML class model is a key element of modern software methodologies and creates in the initial stage of software development. Therefore, error detection and rectification of the UML class model may save software development costs and time. This paper presents an overview of UML Class model verification approaches and identifies open issues, current research trends, and other improvement areas. This study uses a systematic literature review as an investigation method with six research questions and assesses 65 papers dated January 1997 to December 2020. From 2124 published research papers, 65 papers are selected and distributed into 7 studies. This work provides an analysis of verification approaches and the automation level of proposed approaches. As a result, it is found that the existing UML class model verification methods provide great efforts to check correctness. However, in some situations (when dealing with large and complex models), they consume a significant amount of time and do not support many important features of the UML class model.

Knowledge graphs are graph-structured knowledge bases containing abundant entities and relations among the entities. Entity exploration can help users understand the overall structure of knowledge graphs, as well as find the entities of interest in an exploratory manner. Being different from typical entity-centric search whose goal is to retrieve the most related entities for a user’s specific need, the related entities that present diverse aspects are preferred for user’s ambiguous needs in entity exploration. In this paper, we propose a novel diversity-aware entity exploration approach based on random walk model, which naturally mimics human conceptual exploration by surfing a class association graph. This model leverages the diversity, representativeness and relatedness of entity classes to rank these classes in a unified way. Furthermore, for each top ranked class, the associated entities are ranked by combining their diversity and popularity. We compare our algorithm with four baseline algorithms and the experimental results indicate that it outperforms baselines. Furthermore, we conduct a task-based user study to evaluate our approach and the experimental results show that our work provides effective support for entity exploration.

In this paper, an adaptive tracking control scheme is proposed for a class of uncertain systems with external disturbances existing in every state equation and output equation. The output disturbance considered here does not need to be generated by a known different equation. It leads to the system about coordinate changes variables used for virtual controls design no longer meet the triangular requirement of the backstepping approach. Thus the traditional backstepping techniques cannot be used in the controller design. This problem has been solved by the proposed robust adaptive control scheme. In addition, the effect caused by the disturbance in state equation can be compensated for by selecting appropriate design parameters in every step. It is shown that the proposed adaptive control scheme can ensure all signals in the closed-loop system bounded.

Since the advent of distributed ledger technologies, they have provided diverse opportunities in a wide range of application domains. This article brings a comprehensive review of the fundamentals of distributed ledger and its variants. Analyzing 185 publications, ranging from academic journals to industry websites, it provides a comparative analysis of 130 consensus algorithms using a novel architectural classification. The distribution of the reviewed algorithms is analyzed in terms of the proposed classification and different application domains, along with the applicability of each class among the top 10 platforms in the most prominent blockchain application domains. Additional conclusions are drawn from the evolution of consensus mechanisms, and the analysis concludes envisaging future prospects for consensus as an important part of distributed ledger technology.

This study aimed to automatically construct knowledge graphs for online collaborative programming. We proposed several models and developed a system to construct knowledge graphs based on online discussion texts and the target knowledge graph for the C programming language. Our system included two main modules, namely, entity recognition and relation extraction. We proposed an innovative approach for recognizing knowledge entities, which included sequence tagging, text classification, and keyword matching. The extraction of relationships among knowledge entities was performed through queries of the target knowledge graph. The six kinds of knowledge graphs could be automatically generated through our method, including the activated and unactivated knowledge graphs of each student, each group, and each class. The accuracy of entity recognition reached 87.27%. The accuracies of relation extraction for students, groups, and the class reached 89.7%, 90.4%, and 90.2%, respectively. This study is very promising and significant for both teachers and practitioners to provide interventions and personalized learning services based on the constructed knowledge graphs.

Clustering algorithms are a class of unsupervised machine learning (ML) algorithms that feature ubiquitously in modern data science, and play a key role in many learning-based application pipelines. Recently, research in the ML community has pivoted to analyzing the fairness of learning models, including clustering algorithms. Furthermore, research on fair clustering varies widely depending on the choice of clustering algorithm, fairness definitions employed, and other assumptions made regarding models. Despite this, a comprehensive survey of the field does not exist. In this paper, we seek to bridge this gap by categorizing existing research on fair clustering, and discussing possible avenues for future work. Through this survey, we aim to provide researchers with an organized overview of the field, and motivate new and unexplored lines of research regarding fairness in clustering.

The majority of existing few-shot learning methods describe image relations with binary labels. However, such binary relations are insufficient to teach the network complicated real-world relations, due to the lack of decision smoothness. Furthermore, current few-shot learning models capture only the similarity via relation labels, but they are not exposed to class concepts associated with objects, which is likely detrimental to the classification performance due to underutilization of the available class labels. To paraphrase, children learn the concept of tiger from a few of actual examples as well as from comparisons of tiger to other animals. Thus, we hypothesize that in fact both similarity and class concept learning must be occurring simultaneously. With these observations at hand, we study the fundamental problem of simplistic class modeling in current few-shot learning methods. We rethink the relations between class concepts, and propose a novel Absolute-relative Learning paradigm to fully take advantage of label information to refine the image representations and correct the relation understanding in both supervised and unsupervised scenarios. Our proposed paradigm improves the performance of several the state-of-the-art models on publicly available datasets., IEEE/CVF Conference on Computer Vision and Pattern Recognition 2021

We present a visual navigation approach that uses context information to navigate an agent to find and reach a target object. To learn context from the objects present in the scene, we transform visual information into an intermediate representation called context grid which essentially represents how much the object at the location is semantically similar to the target object. As this representation can encode the target object and other objects together, it allows us to navigate an agent in a human-inspired way: the agent will go to the likely place by seeing surrounding context objects in the beginning when the target is not visible and, once the target object comes into sight, it will reach the target quickly. Since context grid does not directly contain visual or semantic feature values that change according to introductions of new objects, such as new instances of the same object with different appearance or an object from a slightly different class, our navigation model generalizes well to unseen scenes/objects. Experimental results show that our approach outperforms previous approaches in navigating in unseen scenes, especially for broad scenes. We also evaluated human performances in the target-driven navigation task and compared with machine learning based navigation approaches including this work.

The training of deep neural networks heavily depends on the data distribution. In particular, these networks easily suffer from class imbalance. The trained networks would recognize the frequent classes better than the infrequent classes. To resolve this problem, existing approaches typically propose novel loss functions to obtain better feature embedding. In this paper, we argue that drawing a better decision boundary is as important as learning better features. Based on our observations, we investigate how the class imbalance affects the decision boundary and deteriorates the performance. We also investigate the feature-distributional discrepancy between training and test time. Accordingly, we propose a novel, yet simple method for class imbalanced learning. Despite its simplicity, our method exhibits outstanding performance. Specifically, the experimental results show that we can significantly improve a network by scaling the weight vectors, even without additional training processes.

Response surface methodology (RSM) refers to experimental designs for optimizing or developing processes, initially in manufacturing. In this paper, a new method is presented and an algorithm is implemented that modifies the axial part in a central composite design to achieve a good D-value and efficiency. The new designs are suitable for sequential experimentation. In comparison with known designs in the same class, the new designs are tested and found to have better D-values on a range of factors. With this new approach, efficient orthogonal designs for response surface methodology were generated for a number of parameters that were previously impossible to construct. The new generated designs and their comparison with known designs from the literature are presented in tables for practitioners' use.

The development of floor cleaning robots is an emerging area in robotics. Maximizing the area coverage is a foremost mission for a floor cleaning robot. Reconfigurable floor cleaning robots outperform floor cleaning robots with fixed morphology in the aspect of area coverage. A reconfigurable robot should be more flexible in changing its morphologies by considering the shapes of objects occupied in an environment to gain more coverage. Nevertheless, the state of the art methods of tiling robots considers only a limited number of morphologies for the reconfiguration, which is not sufficient to match the shape of an object. Therefore, this paper proposes a novel method to synthesize an appropriate morphology for a reconfigurable robot in accordance with the shape of an object. The proposed concept is named hTetro-Infi since it is not limited to a finite number of morphologies. The major novelty of the proposed concept overt the state of the art is the consideration of an infinite number of morphologies for the reconfiguration without sticking into a limited number of morphologies. Feedforward Neural Network (FNN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) were used for determining the hinge angle required for synthesizing a given morphology. Different configurations of FNNs and ANFISs were trained and evaluated to find the most suitable configurations. The area coverage performance of the proposed hTetro-Infi was compared against that of the state of the art methods of an existing class of tiling robots, which considers only a limited number of morphologies, through simulations. According to the statistical conclusions, the proposed hTetro-Infi is capable of significantly improving area coverage compared to an existing tiling-theory based floor cleaning robot. Furthermore, the area coverage improvement of hTetro-Infi is noteworthy. Therefore, the proposed concept is beneficial in improving the abilities of a reconfigurable cleaning robot. Real-world experiments with the hardware platform of the robot for evaluating the performance is expected to be conducted in the next phase of the work. Furthermore, consideration of hTetro-Infi for navigation through confined areas is proposed for future work.

Although a variety of techniques to detect malicious websites have been proposed, it becomes more and more difficult for those methods to provide a satisfying result nowadays. Many malicious websites can still escape detection with various Web spam techniques. In this paper, we first summarize three types of Web spam techniques used by malicious websites, such as redirection spam, hidden IFrame spam, and content hiding spam. We then present a new detection method that adopts the perspective of users and takes screenshots of malicious webpages to invalidate Web spams. The proposed detection method uses a Convolutional Neural Network, which is a class of deep neural networks, as a classification algorithm. In order to verify the effectiveness of the method, two different experiments have been conducted. First, the proposed method was tested based on a constructed complex dataset. We present comparison results between the proposed method and representative machine learning-based detection algorithms. Second, the proposed method was tested to detect malicious websites in a real-world Web environment for three months. These experimental results illustrate that the proposed method has a better performance and is applicable to a practical Web environment.

In this paper, an adaptive backstepping control scheme is proposed for a class of nonlinear systems with unknown time delay. The unknown time delay in the system depends on all system states, which will inevitably destroy the triangular structure of the control system. Therefore, traditional backstepping can not be used to design the controller. In order to overcome the problem caused by “non-triangular structure”, we construct a linear projection between the state vector and its transform in controller design. Using inequality and adaptive backstepping control, we compensate the uncertainties caused by unknown time delays. Finally, the simulation results show that the proposed algorithm is effective.

Frauds and default payments are two major anomalies in credit card transactions. Researchers have been vigorously finding solutions to tackle them and one of the solutions is to use data mining approaches. However, the collected credit card data can be quite a challenge for researchers. This is because of the data characteristics that contain: (i) unbalanced class distribution, and (ii) overlapping of class samples. Both characteristics generally cause low detection rates for the anomalies that are minorities in the data. On top of that, the weakness of general learning algorithms contributes to the difficulties of classifying the anomalies as the algorithms generally bias towards the majority class samples. In this study, we used a Multiple Classifiers System (MCS) on these two data sets: (i) credit card frauds (CCF), and (ii) credit card default payments (CCDP). The MCS employs a sequential decision combination strategy to produce accurate anomaly detection. Our empirical studies show that the MCS outperforms the existing research, particularly in detecting the anomalies that are minorities in these two credit card data sets.

The existing generative Zero-Shot Learning (ZSL) methods only consider the unidirectional alignment from the class semantics to the visual features while ignoring the alignment from the visual features to the class semantics, which fails to construct the visual-semantic interactions well. In this paper, we propose to generate visual features based on an auto-encoder framework paired with multi-modality adversarial networks respectively for visual and semantic modalities to reinforce the visual-semantic interactions with a bidirectional alignment, which ensures the generated visual features to fit the real visual distribution and to be highly related to the semantics. The encoder aims at generating real-like visual features while the decoder forces both the real and the generated visual features to be more related to the class semantics. To further capture the discriminative information of the generated visual features, both the real and generated visual features are forced to be classified into the correct classes via a classification network. Experimental results on four benchmark datasets show that the proposed approach is particularly competitive on both the traditional ZSL and the generalized ZSL tasks.

In the era of the fourth industrial revolution (Industry 4.0) and the Internet of Things (IoT), real-time data is enormously collected and analyzed from mechanical equipment. By classifying and characterizing the measured signals, the fault condition of mechanical components could be identified. However, most current health monitoring techniques utilize time-consuming and labor-intensive feature engineering, i.e., feature extraction and selection, that are carried out by experts. This paper, on the contrary, deals with an automatic diagnosis method of machine monitoring using a convolutional neural network (CNN) with class activation maps (CAM). A class activation map enables us to discriminate the fault region in the images, thus allowing us to localize the fault precisely. The goal of the paper is to demonstrate how CNN and CAM could be employed to real-world vibration video to characterize the machine’s status, representing normal or fault conditions. The performance of the proposed model is validated with a base-excited cantilever beam dataset and a water pump dataset. This paper presents a novel industrial application by developing a promising method for automatic machine condition-based monitoring.

Three-way decision (3WD) provides a new perspective for solving practical decision-making problems, which is in line with human's cognitive pattern. A covering information system (CIS) is an information system (IS) that consists of multiple coverings in the universe. A CIS with decision attributes which is seen as a covering decision information system (CDIS). This paper proposes three-way group decisions in a CDIS, as well as gives its application on the problem of position competition. First of all, the neighbourhood of every point in a CDIS is defined, and corresponding similarity class of this point is also obtained. Then, because of the uncertainty of risks, loss functions are acquired through group decision-making by means of interval numbers. Next, a method of three-way group decisions in a CDIS is presented. Eventually, the position competition is presented as an example to support our proposed decision-making method.

Despite the recent great achievements in speech emotion recognition (SER) with the development of deep learning, the performance of SER systems depends strongly on the amount of labeled data available for training. Obtaining sufficient annotated data, however, is often extremely time consuming and costly and sometimes even prohibitive because of privacy and ethical concerns. To address this issue, this article proposes the semisupervised generative adversarial network (SSGAN) for SER to capture underlying knowledge from both labeled and unlabeled data. The SSGAN is derived from a GAN, but the discriminator of the SSGAN can not only classify its input samples as real or fake but also distinguish their emotional class if they are real. Thus, the distribution of realistic inputs can be learned to encourage label information sharing between labeled and unlabeled data. This article proposes two advanced methods, i.e., the smoothed SSGAN (SSSGAN) and the virtual smoothed SSGAN (VSSSGAN), which, respectively, smooth the data distribution of the SSGAN via adversarial training (AT) and virtual adversarial training (VAT). The SSSGAN smooths the conditional label distribution given inputs using labeled examples, while the VSSSGAN smooths the conditional label distribution without label information (“virtual” labels). To evaluate the effectiveness of the proposed methods, four publicly available and frequently used corpora are selected to conduct experiments in intradomain and interdomain situations. The results illustrate that the proposed methods are superior to the state-of-the-art methods. Specifically, in experimental settings with mismatched and semimismatched unlabeled training sets, the SSSGAN and VSSSGAN are more robust than the SSGAN because of the distributional smoothness.

In the era of web 2.0, online forums, blogs and Twitter are becoming primary sources for sharing views, opinions and comments about different topics. Classifying these views, opinions and comments is known as sentiment analysis which is an active research area. Sentiment analysis has vast applications in different fields of life, such as marketing, e-commerce and business. Under the umbrella of sentiment analysis, sentiment quantification that deals with estimating relative frequency of class of interest is being investigated by researchers nowadays. In sentiment quantification, exploring effect of new features and comparison of diverse types of classifiers to assess their effectiveness needs further investigation. In this paper, we explore diverse feature sets and classifiers for sentiment quantification. In addition, empirical performance analysis of conventional machine learning techniques, ensemble-based methods and state-of-the-art deep learning algorithms on basis of features set, is performed. The computed results show that the diverse features sets affect the performance of classifiers in sentiment quantification. The results also confirm that the deep learning techniques perform better than the conventional machine learning algorithms.

This paper concerns a novel data-based policy iteration control method for discrete-time systems with disturbances. To avoid the need of system dynamic and state, a new system equation is constructed by using system operation data. Based on this, we derive a new data-based Bellman equation that can be applied to the design of model-free policy iteration algorithm. For discrete-time system with disturbances, we transform H∞ control into a zero-sum game problem, and then use the proposed new data-based PI method to solve it. Wind turbine control is a typical control problem with time delay and disturbance. Mathematical simulation results are given to demonstrate the effectiveness of the proposed data-based PI control method.

With the advent of the era of big data in the world and the commercial value of face recognition technology, the prospects for face recognition technology are very bright and have great market demand. This article aims to design a face recognition attendance system based on real-time video processing. This article mainly sets four directions to consider the problems: the accuracy rate of the face recognition system in the actual check-in, the stability of the face recognition attendance system with real-time video processing, the truancy rate of the face recognition attendance system with real-time video processing and the interface settings of the face recognition attendance system using real-time video processing. By analyzing the situation of these problems, the concept of attendance system based on face recognition technology is proposed, and the research on face recognition attendance system based on real-time video processing is carried out. Experimental data shows that the accuracy rate of the video face recognition system is up to 82%. Compared with the traditional check-in method, the face recognition attendance system can be reduced by about 60%. The rate of skipping classes has greatly reduced the phenomenon of students leaving early and skipping classes. The face recognition time and attendance system with real-time video processing through the above experimental certification can quickly complete the tasks of students in the time and attendance check-in system, get rid of the complex naming phenomenon, greatly improve the efficiency of class, and play an important role in guiding the development of the time and attendance system.

Role-based access control (RBAC) is the most popular access control model currently adopted in several contexts to define security management. Constraints play a crucial role since they can drive the selection of the best representation of the organization's security policies when migrating towards an RBAC system. In this paper, we examine different types of constraints addressing both theoretical aspects and practical considerations. On one side, we define the constrained role mining problem for each constraint type, showing its complexity. On the other hand, we present efficient heuristics adapted to each class of constraints, all derived from the specialization of a general approach for role mining. We show that our techniques improve over previous proposals, offering a complete set of experimentations obtained after the application of the heuristics to standard real-world datasets.

Hashing based methods for cross-modal retrieval has been widely explored in recent years. However, most of them mainly focus on the preservation of neighborhood relationship and label consistency, while ignore the proximity of neighbors and proximity of classes, which degrades the discrimination of hash codes. And most of them learn hash codes and hashing functions simultaneously, which limits the flexibility of algorithms. To address these issues, in this article, we propose a two-step cross-modal retrieval method named Manifold-Embedded Semantic Hashing (MESH). It exploits Local Linear Embedding to model the neighborhood proximity and uses class semantic embeddings to consider the proximity of classes. By so doing, MESH can not only extract the manifold structure in different modalities, but also can embed the class semantic information into hash codes to further improve the discrimination of learned hash codes. Moreover, the two-step scheme makes MESH flexible to various hashing functions. Extensive experimental results on three datasets show that MESH is superior to 10 state-of-the-art cross-modal hashing methods. Moreover, MESH also demonstrates superiority on deep features compared with the deep cross-modal hashing method.

Zero Shot Learning (ZSL) has been attracting increasing attention due to its powerful ability of recognizing objects of unseen classes. As one type of ZSL methods, the low rank based strategy has achieved remarkable success. However, traditional low rank based methods are often based on the assumption that a variety of visual features from a same class can be projected to a single attribute by ignoring the background information and other noisy interference in visual features. This assumption is unreasonable and often leads to bad performance when there is big variance within a class. In this paper, a novel method called Prototype Relaxation with Robust Principal Component Analysis (RPCA) is proposed to relax this assumption by adding a sparse noise constraint. In addition, to avoid the confusion between similar classes, an orthogonal constraint is employed to disperse all the class prototypes, including both seen and unseen classes, in latent space. Furthermore, to alleviate the domain shift problem, vectors from latent space are exploited to reconstruct visual features and semantic attributes respectively. Besides, the hubness problem is also mitigated by applying the max probability model in all three spaces. Extensive experiments are conducted on four popular datasets and the results demonstrate the superiority of this method.

This paper proposes an event-triggered fuzzy adaptive control for a class of uncertain nonlinear systems with input dead-zone. Considering that the approximation error is a time-varying function, improved Fuzzy Logic Systems(FLSs) are applied to address the uncertain parts of the considered systems. And a designed auxiliary system is introduced into the control design to compensate the effects of input dead-zone, which often occurs to actual systems. Moreover, a novel event-triggered strategy is established to further save system transmission resources. It contains a two-bits signal transmission mechanism. According to the theoretical analysis and simulation results, the boundness of all signals can be guaranteed and the validity of the proposed method can be verified.

As eating-out became an indispensable part of our daily lives, demand for the food recognition of unfamiliar restaurant increased significantly due to health-care. Although there are many researches on generic food recognition, there are relatively fewer studies on restaurant food image recognition. Meanwhile, it becomes extremely challenging for restaurant food image recognition due to insufficient food image. Prototypical network is common utilized to address the such a task in recent years. Although the methods based on prototypical network achieve impressive results in capturing similarities feature of the same food category, it fails to highlight important information on feature and instance level. Toward this end, we propose an effective food image recognition scheme by incorporating hybrid attention mechanism into prototypical network in this paper. Specifically, the image feature is first captured by convolutional neural network (CNN). Then the image attention weights yielded by instance-based attention mechanism are used to modulate the image feature of CNN for constructing class prototypes. And feature-based attention mechanism is employed to grasp important information of image for enriching image representation. Extensive experimental results on the large food image dataset verify that the performance of our proposed classification scheme outperforms the state-of-the-art ones.

Visual attention mechanism has been widely used in computer vision and plays a vital role in the research of human action recognition. In this paper, we explore a novel moving target detection mechanism for human action recognition and propose a new 3D CNN (3D Convolutional Neural Network) model, dubbed LaM-2SRN (Local Features Enhanced and Moving target detected 2Stream-ResNet) for extracting and learning attention-enhanced spatiotemporal features. The contributions of this paper are as follow: First, the traditional CAM (Class Activation Map) based visual attention algorithm is used to obtain the optical flow information of the human region, thus eliminating the influence of irrelevant optical flow information (such as background clutter). Second, the ViBe algorithm is used to identify the moving target in the continuous frame, retain the optical flow information of the moving target, and make it complement with the optical flow information of the human region, to obtain a complete motion descriptor. After the motion information is marked in the video frame, we put the marked video frames and the original video frames into the 3D CNN and 2D CNN models respectively, to acquired mixed descriptor, which serves as the basis for video classification. Different from most of the previous target detection algorithms, this paper not only detects the human target, but also detects the moving foreground targets in the video, so that the pre-trained CNN model can obtain more complete motion information from videos. We use RGB-only video data for evaluation on two benchmarks: UCF101 and HMDB51, the experimental results demonstrate that our LaM-2SRN is comparable to the previous state-of-the-art algorithms.

Median Access Control (MAC) protocols are designed to mitigate collisions and enhance the energy efficiency for sensor data collection. This paper reviews two basic categories of MAC protocols. The first class is the contention-based protocols, where nodes randomly compete for channel access. The second class is the schedule-based MAC protocols, in which nodes access the channel on the basis of the predetermined schedules. We focus on discussing the Time Division Multiple Access (TDMA) protocols and classify different TDMA schedulings into three categories according to the communication patterns in the network, i.e., the link scheduling, aggregate scheduling and non-aggregate scheduling. Link scheduling deals with the peer to peer communication pattern, where there is no central node in the network. In comparison, both the aggregate and non-aggregate schedulings handle the convergecast communication pattern, in which all traffic are destined to the sink. This survey provides a comprehensive overview of these three categories and provided a detailed briefing on how the TDMA schedules handle the network traffic dynamics in the network. Compared with other surveys in this domain, this review does not confine itself to deal with collecting a particular form of data, but provides a unified framework to integrate the data semantics in a broader sense into the design of TDMA scheduling algorithms.

Semantic segmentation is one of the most commonly used techniques for road scene understanding. Recently developed deep learning-based semantic segmentation networks are typically based on the encoder-decoder structure and have made great progress in road scene understanding. However, these conventional networks still encounter difficulties in recovering spatial details. To overcome this problem, we introduce a lightweight prediction and boundary-aware refinement module that can hierarchically refine the segmentation results with spatial details. The proposed refinement module has two attention units called the upper-level prediction attention unit and the upper-level boundary attention unit. The upper-level prediction attention unit emphasizes the features in the regions that need to be refined by using predicted class probability from the upper-level, whereas the upper-level boundary attention unit focuses on the features near the semantic boundary of the upper-level segmentation result. By using the proposed prediction and boundary-aware refinement module in the decoder network, the segmentation result can gradually be improved in a top-down manner to a finer and more complete one. Experimental results on the Cityscapes and CamVid datasets demonstrate that the proposed prediction and boundary attention-based refinement module can achieve considerable performance improvement in segmentation accuracy with a marginal increase in computational complexity.

Unmanned aerial vehicles (UAVs) also known as drones are increasingly populating our skies. This represents a relevant issue both for the legislator and the researcher. While the regulation plans often assume precautionary approaches, stating restrictive conditions of use for the sake of public safety, the applied research is exploring novel strategies to develop autonomous vehicles endowed with trusty operating mechanisms. The challenge is to let drones overflying even populated areas while keeping a steady control of the situation on the ground, thus enabling the possibility of safe landing with no harm for people. This can be done employing on-board cameras and embedded GPUs which allow for the execution in real-time of computer vision applications. In this paper, we introduce a crowd detection method for drone safe landing. The pivotal points of our proposal are related to the computational limitations imposed by the currently available hardware resources of UAVs. In this sense, our method is based on the light-weight scheme of a fully-convolutional neural network which conjugates nimble computations and effectiveness. We propose a two-loss model where a classification task (oriented to distinguish between crowded/non-crowded scenes) is supported by a regression task (aimed at better focusing the agglomeration tendency of the persons). This latter job is realized by resorting to the construction of a spatial graph for each analysed image and to the evaluation of the corresponding clustering coefficient. As a further element, our model is endowed with the capability to produce class activation heatmaps which contribute to the semantic enrichment of the flight maps. We tested our model on a large dataset of aerial images and we observed how it compares favorably with other approaches proposed in literature.

Since its inception in 2015, Style Transfer has focused on texturing a content image using an art exemplar. Recently, the geometric changes that artists make have been acknowledged as an important component of style[42], [55], [62], [63]. Our contribution is to propose a neural network that, uniquely, learns a mapping from a 4D array of inter-feature distances to a non-parametric 2D warp field. The system is generic in not being limited by semantic class, a single learned model will suffice; all examples in this paper are output from one model.Our approach combines the benefits of the high speed of Liu et al. [42] with the non-parametric warping of Kim et al. [55]. Furthermore, our system extends the normal NST paradigm: although it can be used with a single exemplar, we also allow two style exemplars: one for texture and another geometry. This supports far greater flexibility in use cases than single exemplars can provide.

Digitalization of manufacturing is led by the notion of a digital thread, a framework that allows for integrating authoritative and trustable data of a product throughout its lifecycle. Under the model-based enterprise (MBE) vision, digital three-dimensional (3D) model of a product serve as this authoritative source of information across the product lifecycle. A class of systems called knowledge-based engineering (KBE) systems represents the evolution of knowledge-based systems towards designing these models and this study presents a case for using such systems to realize the digital thread in a model-based enterprise. The paper proposes four directions and potential enabling technologies at the disposal of KBE systems for this purpose are identified. Further, a case study that demonstrates such use of a KBE system to realize the digital thread for a manufacturing assembly process is presented. publishedVersion

This work proposes SpecRank, an algorithm for the ranking of objects from pairwise comparisons provided by workers with different specialties in a crowdsourced setting. We assume that the objects may belong to different latent classes that are unknown to the requester, and the workers' ability to provide reliable labels may depend on their specialty or familiarity with these classes. We first propose a specialty Bradley-Terry-Luce (specialty-BTL) model that incorporates workers' specialties in the quality of their pairwise comparison labels. Moreover, we adopt a latent feature model where the scores of objects are modeled as inner products between the latent feature vectors of the objects and their associated classes. The workers' ability to provide accurate pairwise labels is also determined by the correlation between their specialty feature vectors and the class feature vectors. By adopting a mixture model on the objects' class labels, our proposed SpecRank adopts variational expectation maximization (EM) algorithm to perform joint ranking and clustering of the objects. The effectiveness of the proposed specialty-BTL model and the SpecRank algorithm is demonstrated on both synthetic and real-world datasets.

With the rapid development of Internet technology and intelligent technology, it is an inevitable trend for manufacturing industry to move from traditional manufacturing to digital manufacturing and intelligent manufacturing. As the "mother machine" of production and manufacturing, CNC equipment has various types and communication modes, which makes it difficult to monitor various types of CNC equipment in the workshop. This paper proposes an abstract connection method for CNC machine tools by comparing and analyzing the common communication methods and data acquisition schemes of various mainstream CNC systems. It can be applied to the data acquisition of most advanced CNC machine tools, and can effectively solve the communication obstacles between enterprise management system and different CNC systems. The code of machine tools connected by the server is reduced to at least 20% of the original.

Malware is growing rapidly in number and become more and more sophisticated. To prevent them we need to collect samples continuously and update them to the classifier. In this paper, we will propose a method to update new labeled samples of malware to the classifier easily without re-training everything. The classifier can be updated by both labeled malware samples of an existing class or a new class. Our method also has the ability to detect samples of unknown families. Experiments are performed over the traditional computer malware dataset and the IoT malware dataset. The results have shown that our method can reach the macro F1-score almost the same re-train everything but take significantly less time.

Sentiment analysis is an important area of natural language processing (NLP) research, and is increasingly being performed by machine learning models. Much of the work in this area is concentrated on extracting sentiment from textual data sources. Clearly however, a textual source does not convey the pitch, prosody, or power of the spoken sentiment, making it attractive to extract sentiments from an audio stream. A fundamental prerequisite for sentiment analysis on audio streams is the availability of reliable acoustic representation of sentiment, appropriately labeled. The lack of an existing, large-scale dataset in this form forces researchers to curate audio datasets from a variety of sources, often by manually labeling the audio corpus. However, this approach is inherently subjective. What appears “positive” to one human listener may appear “neutral” to another. Such challenges yield sub-optimal datasets that are often class imbalanced, and the inevitable biases present in the labeling process can permeate these models in problematic ways. To mitigate these disadvantages, we propose the use of a text-to-speech (TTS) engine to generate labeled synthetic voice samples rendered in one of three sentiments: positive, negative, or neutral. The advantage of using a TTS engine is that it can be abstracted as a function that generates an infinite set of labeled samples, on which a sentiment detection model can be trained. We investigate, in particular, the extent to which such training exhibits acceptable accuracy when the induced model is tested on a separate, independent and identically distributed speech source (i.e., the test dataset is not drawn from the same distribution as the training dataset). Our results indicate that this approach shows promise and the induced model does not suffer from underspecification.

This manuscript reports the results of an experiment to assess the effectiveness of multi-abstraction computing tools on promoting exploratory self-learning in engineering, using a custom Real-Time Operating System (RTOS) during remote learning. Our goals were to determine whether students respond to guided exploratory learning opportunities during remote learning, and whether these contributed to immediate and further student success.Students in a 3rd year course were given the custom multi-abstraction RTOS. The higher abstraction layer was taught directly, whilst students were directed towards exploratory self-learning among the lower abstraction layer (which includes details beyond prior knowledge or the content of the course). In a survey, students were asked about their perceptions of using higher and lower abstraction layers and how these impacted their learning. Responses were correlated with course grades.Students that engaged with the lower abstraction layer demonstrated higher success (3 percentage points above class average), engagement (22.5% reporting they were more engaged), accuracy of self-assessment (1.3 percentage points above class average), and learning beyond course’s expectations (14% reporting they learned beyond the course’s learning outcomes).

COVID 19 pandemic caused by SARS-CoV-2, has managed to deepen and to put on evidence, the great inequalities existing between countries such as the internal social strata regarding accessibility to education and technological services that are required to have effective distance learning. Although rural areas still have precarious basic services throughout the continent, it is enough with a single non-studying child to conclude that there is no democratization of learning access and technology. The research methodology was a bibliographic review, supported on electronic means to obtain all the relevant and available information about the subject of study. The general conclusion is that regardless of the accessibility to learning that exist in our countries, the important thing is the knowledge that remains, if this is really effective, if the teacher do the class online with the available applications, and also if he has all the technological resources, to teach the class and not only to send the weekly planning of the assignment either by mail, messaging or web platforms, without an evaluation to verify that the student has learned and passed the class or subject

One approach to teaching game design to students with a wide variety of disciplinary backgrounds is through team game projects that span multiple weeks, up to an entire term. However, open-ended, creative projects introduce a gamut of challenges to novice programmers. Our goal is to assist game design students with the planning stage of their projects. This paper describes our data collection process through three course interventions and student interviews, and subsequent analysis in which we learned students had difficulty expressing their creative vision and connecting the game mechanics to the intended player experience. We present these results as a step towards the goal of scaffolding the planning process for student game projects, supporting more creative ideas, clearer communication among team members, and a stronger understanding of human-centered design in software development.

Recent years have seen the rapid adoption of artificial intelligence (AI) in every facet of society. The ubiquity of AI has led to an increasing demand to integrate AI learning experiences into K-12 education. Early learning experiences incorporating AI concepts and practices are critical for students to better understand, evaluate, and utilize AI technologies. AI planning is an important class of AI technologies in which an AI-driven agent utilizes the structure of a problem to construct plans of actions to perform a task. Although a growing number of efforts have explored promoting AI education for K-12 learners, limited work has investigated effective and engaging approaches for delivering AI learning experiences to elementary students. In this paper, we propose a visual interface to enable upper elementary students (grades 3–5, ages 8–11) to formulate AI planning tasks within a game-based learning environment. We present our approach to designing the visual interface as well as how the AI planning tasks are embedded within narrative-centered gameplay structured around a Use-Modify-Create scaffolding progression. Further, we present results from a qualitative study of upper elementary students using the visual interface. We discuss how the Use-Modify-Create approach supported student learning as well as discuss the misconceptions and usability issues students encountered while using the visual interface to formulate AI planning tasks.