Search

• A UCI online news popularity dataset was utilized in this study, and the quality of the dataset was improved after using data pre-processing methods such as normalization and principal component analysis. The number of shares was transformed into a popularity or unpopularity classification problem. • The experimental results show that using a single-class classification algorithm is better than a two-class classification algorithm when the data is unbalanced. • The contribution of this paper is to modify the classification boundaries from 1400 to 50,000 in order to form imblanced data distributions and attempt to combine an Autoencoder and One-Class SVM algorithms to overcome the problem of imbalanced data. Due to its fast transmission and easy accessibility features, the Internet has replaced traditional newspapers and magazines as the main channel for delivering public news. Hence, predicting the popularity of Internet news has become an essential topic. This research is based on a UCI dataset, the primary source of which is Mashable News, one of the major blogs in the world. The number of shared articles is used as a predictor of the popularity of the news, and the four types of machine learning algorithms utilized are Random Forest, LightGBM, XGBoost, and One-Class SVM. The best prediction method is One-Class SVM with 88% accuracy. This result indicates that combining Autoencoder and One-Class algorithm will optimize the prediction while detecting anomalies within imbalanced data. One-class SVM algorithm based on an autoencoder adopted in this study outperforms other algorithms, namely Random Forest, XGBoost, and LightGBM, in the category of the accuracy, the precision, the recall, and F1 scores. [Display omitted] [ABSTRACT FROM AUTHOR]

• Security of information in internet of medical things systems is paramount. • Authentication mechanism is enhanced by machine learning and blockchain. • Smart contracts verify the valid access to information or revoke it otherwise. • Computational time is reduced when machine learning algorithm is used. Due to the fast-pacing development of technology in the healthcare domain, many problems arise surrounding the security and privacy preservation of medical data. Secure authentication on the Internet of Medical Things (IoMT) is essential. The lack of security in critical and sensitive information of IoMT may lead to high-risk issues in patient privacy. When new data is transmitted from the sensor node, it cannot be assured as authenticated data. Therefore, a blockchain-based system is needed. Such a system allows healthcare providers to access the health records of patients in a more secured authentication-based approach across various network connections. In this paper, a new secure authentication approach using machine learning is proposed. To identify the dynamic time attack detection and authentication in an IoMT environment, this work implements K-Nearest neighbour (KNN) and machine learning using smart contract (KNN-MLSC). It improves security, reduces latency, and maintains health data privacy for both physicians and patients. The accuracy of KNN-MLSC got 0.96 compared with KNN using a smart contract. Also, the results showed that KNN-MLSC has the minimum computation time. [Display omitted] [ABSTRACT FROM AUTHOR]

Autonomous vehicles are the latest engineering research sensations. The development of these needs the collaboration of studies in the automotive and artificial intelligence domains. Self-driving cars face a slew of security issues when left on their own without human supervision. However, the massive volumes of data required for vehicle operation can't be processed rationally by humans. This is where modern intelligence systems can help. They can facilitate clear decision-making regarding the state of the systems. A unique technique based on Machine Learning (ML) procedures is proposed in this paper. Two cluster models have been constructed and coupled with three Convolutional Neural Networks (CNN) models based on Demand-Side Management (DSM). These models are integrated to determine the optimal operating time, stability, and management model. Data from the previous three years have been used in this study. The results show that the proposed CNN and cluster 2 models are the best for energy optimization. The best thing is that this real-time model can meet the demands of large traffic. [Display omitted] [ABSTRACT FROM AUTHOR]

• The main objective of this paper is to improve the performance of un-completed face recognition by hybrid-supervised method. • HMMFA is proposed, which improved the inter-class separability and the efficiency of feature extraction. • Image partition is performed before PCANet to enhance its robustness for occlusion. It is still an important and challenging problem for face recognition with occlusion, small sample size, various expressions, and poses, called un-completed face recognition. So we design a simple but effective hybrid-supervision learning frame by fusing the advantages of supervised and unsupervised features. In the supervised branch, we propose an effective feature learning method: HMMFA. In the unsupervised branch, we improve the PCANet to extract more effective local information. In the fusion stage, we further extract the discriminant features contained in the hybrid features and then take SVM as the final classifier. Because the proposed method requires no auxiliary set and has less parameter number than that of deep learning methods, it has a low storage requirement, which makes it more economical and practical for small communities. Experiments on four databases show the effectiveness and efficiency of our method. [Display omitted] [ABSTRACT FROM AUTHOR]

The counts of various types of white blood cells give vital information for identifying a variety of ailments utilizing the video internet of things (VIoT). This technique needs to be automated to save time and minimize counting errors. This paper aims to apply recent image processing techniques to identify and categorize white blood cells (WBCs) found in peripheral blood and to develop a system that would automatically detect and evaluate WBCs. The proposed method consists of steps such as segmentation scanning, feature extraction, and blood cell categorization. First, we segment the cell pictures, which entails grouping white blood cells into clusters. The second stage consists of scanning each input image and preparing the dataset. The third stage involves determining the texture and contour of a scanned picture. Finally, different machine learning algorithms are used to classify the results based on these criteria. [ABSTRACT FROM AUTHOR]

Zero-shot learning is one of the most challenging machine learning tasks, in which learning stable and transferable knowledge from seen classes plays a pivotal role. To improve the currently unsatisfactory performance of zero-shot object recognition, this paper proposes a novel image representation method, namely, micro-knowledge. In our method, the segmentation of micro-regions and the consequent learning of micro-knowledge are unified by the introduction of a self-attention mechanism. A zero-shot classification framework is carefully designed based on micro-knowledge of images. Under this framework, multiple micro-region descriptions are first obtained by embedding micro-knowledge and then merged to carry out the final classification of unseen objects. Finally, a capsule-unified framework is employed as a graphical programming tool to accomplish the aforementioned tasks. Experiments on public datasets show that the proposed framework can generally achieve competitive results for the classification of unseen objects. Specifically, these results verify that the micro-knowledge learned from one dataset can be directly applied to others without complicated adjustments and demonstrate that using visual features instead of semantic features can result in a decrease in classification error. This research will bring new ideas into the field of zero-shot learning and will serve as an appealing option when addressing the problem of domain shift. [ABSTRACT FROM AUTHOR]

• The main objective of this research is to detect Covid -19 virus through machine learning • CT-Lung screening and CAD Diagnosis is performed for Image Preprocessing • Radial basis function and SVM Classifiers have been used for autonomous detection of infected areas As people all over the world are vulnerable to be affected by the COVID-19 virus, the automatic detection of such a virus is an important concern. The paper aims to detect and classify corona virus using machine learning. To spot and identify corona virus in CT-Lung screening and Computer-Aided diagnosis (CAD) system is projected to distinguish and classifies the COVID-19. By utilizing the clinical specimens obtained from the corona-infected patients with the help of some machine learning techniques like Decision Tree, Support Vector Machine, K-means clustering, and Radial Basis Function. While some specialists believe that the RT-PCR test is the best option for diagnosing Covid-19 patients, others believe that CT scans of the lungs can be more accurate in diagnosing corona virus infection, as well as being less expensive than the PCR test. The clinical specimens include serum specimens, respiratory secretions, and whole blood specimens. Overall, 15 factors are measured from these specimens as the result of the previous clinical examinations. The proposed CAD system consists of four phases starting with the CT lungs screening collection, followed by a pre-processing stage to enhance the appearance of the ground glass opacities (GGOs) nodules as they originally lock hazy with fainting contrast. A modified K-means algorithm will be used to detect and segment these regions. Finally, the use of detected, infected areas that obtained in the detection phase with a scale of 50×50 and perform segmentation of the solid false positives that seem to be GGOs as inputs and targets for the machine learning classifiers, here a support vector machine (SVM) and Radial basis function (RBF) has been utilized. Moreover, a GUI application is developed which avoids the confusion of the doctors for getting the exact results by giving the 15 input factors obtained from the clinical specimens. [Display omitted] [ABSTRACT FROM AUTHOR]

Integrated Intelligent Transportation System (IITS) is built by the emergence of the Internet of Vehicles (IoV) and the employment of the Internet of Things (IoT) in the transportation field for improving traffic efficiency, avoiding accidents, and improving the safety experience in driving. Large-scale networks, dynamic topology structure, uneven nodes distribution, and unlimited mobility are the characteristics of IITS. It increased the risk of a variety of attacks related to authentication, credentials, availability, privacy, routing, and legitimacy of data. Distributed Denial of Service (DDoS) flooding attack plays a vital role in impairing Road Side Units (RSU) of IITS and is capable of evading classical data filtering methods. In this paper, a novel DDoS flooding attack detection method is proposed for ensuring the security of IITS by employing reinforcement learning. The proposed framework is implemented by addressing the security requirements of ITS and overcoming the flaws of existing detection methods in advanced vehicular systems. In addition, the Q learning algorithm is designed along with the framework. Other attacks, in addition to DDoS flooding attacks, should be mitigated in the future to ensure a completely secure connection. [ABSTRACT FROM AUTHOR]

While deep learning models are thought to be resistant to random perturbations, it has been demonstrated that these architectures are vulnerable to deliberately crafted perturbations, albeit being quasi-imperceptible. These vulnerabilities make it challenging to deploy Deep Neural Network (DNN) models in security-critical areas. Recently, many research studies have been conducted to develop defense techniques enabling more robust models. In this paper, we target detecting adversarial samples by differentiating them from their clean equivalents. We investigate various metrics for detecting adversarial samples. We first leverage moment-based predictive uncertainty estimates of DNN classifiers derived through Monte-Carlo (MC) Dropout Sampling. We also introduce a new method that operates in the subspace of deep features obtained by the model. We verified the effectiveness of our approach on different datasets. Our experiments show that these approaches complement each other, and combined usage of all metrics yields 99 % ROC-AUC adversarial detection score for well-known attack algorithms. [ABSTRACT FROM AUTHOR]

• A novel deep learning models for arrhythmia detection in IoT healthcare applications are proposed. • It has the independence of the ECG signal quality, as operation is on the spectrograms of the ECG signals. • Ability to work on a single ECG lead, which reduces the complexity of operations. • Applicability of the proposed framework in the smart healthcare platform for real-time operation. • The proposed framework overcomes the overfitting problems and achieves high accuracy on several datasets. In this paper, novel convolutional neural network (CNN) and convolutional long short-term (ConvLSTM) deep learning models (DLMs) are presented for automatic detection of arrhythmia for IoT applications. The input ECG signals are represented in 2D format, and then the obtained images are fed into the proposed DLMs for classification. This helps to overcome most of the problems of the previous machine and deep learning models such as overfitting, and working on more than one lead of ECG signals. We use several publicly available datasets from PhysioNet such as MIT-BIH, PhysioNet 2016 and PhysioNet 2018 for model assessment. Overall accuracies of 97%, 98 %, 94 % and 91 % are obtained on spectrograms of MIT-BIH dataset, compressed MIT-BIH dataset, PhysioNet 2016 dataset, and PhysioNet 2018 dataset, respectively. Compared to the previous works, the proposed framework is more robust and efficient, especially in the case of noisy data. Deep Learning Models for Arrhythmia Detection in IoT Healthcare Applications. [Display omitted] [ABSTRACT FROM AUTHOR]

In clinical documents, privacy and confidentiality protection are the two main challenges before sharing or publishing data. According to the Health Insurance Portability and Accountability Act (HIPAA) and the General Data Protection Regulation (GDPR), even a few terms can cause privacy threats. In retrospect, confidentiality threats are not fully explored due to the complex nature as well as massive number of clinical terms and phrases. Current approaches use information theoretic-based techniques to detect and sanitize risky semantically-correlated terms. However, they have language ambiguity and non-monotonic behavior, coupled with the fact that pre-trained classifiers and human-tagging are required to construct classifiers. This paper offers a generic and adaptable method for protecting risky terms in clinical data using word embedding (Word2Vec and BERT) for risky term detection and comparative analysis. Our methodology uses WordNet taxonomy to minimize a document's semantic and utility loss by substituting privacy-preserving generalization for disclosive words and by eliminating manual data tagging. The results show significant protection and utility-preservation, compared to information-theoretic approaches. [ABSTRACT FROM AUTHOR]

• Empirical Mode Decomposition technique is used for feature extraction to detect power quality disturbances. • Kriging interpolation-based Empirical Mode Decomposition enhances the performance. • Extreme Learning Machine is used for the classification of power quality disturbances. • The robustness of the Extreme Learning Machine is enhanced by the Symbiotic Organism Search Optimization technique. • Validated the efficacy of the proposed approach on the hardware experimental setup. In this paper, a novel approach based on Empirical Mode Decomposition (EMD) and an Extreme Learning Machine (ELM) for the detection and classification of Power Quality Events (PQEs) is proposed. The EMD technique is used for computing the prominent features required to characterize the PQE signals. A down-sampled Kriging Interpolation (KI) based EMD is suggested to enhance the performance of the EMD operation in terms of accuracy and speed. The ELM is applied for the classification of Power Quality Disturbances (PQDs), considering all the derived features through the KI-EMD approach. Symbiotic Organism Search (SOS) optimization technique is applied to enhance the performance and robustness of ELM by optimally computing the values of the system parameters. The performance of the proposed approach is justified with test cases under diverse noise conditions. Comparative results and analysis are provided to show an improvement of 2-5% in terms of accuracy, speed, and robustness compared to other conventional methods. Experimental results validate the efficacy of the proposed approach under real-time conditions. [Display omitted] [ABSTRACT FROM AUTHOR]

Today, incorporating advanced machine learning techniques into intrusion detection systems (IDSs) plays a crucial role in securing mobile edge computing systems. However, the mobility demands of our modern society require more advanced IDSs to make a good trade-off between coping with the rapid growth of traffic data and responding to attacks. Thus, in this paper, we propose a lightweight distributed IDS that exploits the advantages of centralized platforms to train and learn from large amounts of data. We investigate the benefits of two promising bio-inspired optimization algorithms, namely Ant Lion Optimization and Ant Colony Optimization, to find the optimal data representation for the classification process. We use Deep Forest as a classifier to detect intrusive actions more robustly and generate as few false positives as possible. The experiment results show that the proposed approach can enhance the reliability of lightweight intrusion detection systems in terms of accuracy and execution time. [Display omitted] • Hybrid learning model has benefits in supporting intrusion detection in mobile edge computing. • Deep Forest is effective in strengthening attack detection predictability of IDSs in mobile edge computing. • Bio-inspired optimization algorithms improve the classification process. • The employed strategy enhances attack recognition of IDSs to deal with different threats in mobile edge computing. [ABSTRACT FROM AUTHOR]

• Develop a deep learning model for automated small ship detection and classification. • Employ hyperparameter tuned Mask RCNN with SqueezeNet model for ship detection. • Introduce an optimal weighted regularized ELM model for small ship classification. Autonomous ship technologies have gained considerable interest due to the minimization of the challenging issues faced by the unpredictable errors of manual navigation, and therefore reduces human labor, increasing navigation security and profit margin. On autonomous shipping technologies, small ship detection is vital in ensuring shipping safety. With this motivation, this paper presents an efficient optimal mask regional convolutional neural network (Mask-CNN) technique for small ship detection (OMRCNN-SHD) on autonomous shipping technologies. Primarily, the data augmentation process is performed to resolve the issue of the limited number of real-world samples of small ships and helps to detect small ships in most cases accurately. Besides, the Mask RCNN with SqueezeNet model is used to detect ships and the hyperparameter tuning of the SqueezeNet model takes place by the use of the Adagrad optimizer. Furthermore, the Colliding Body's Optimization (CBO) algorithm with the weighted regularized extreme learning machine (WRELM) technique is employed to classify detected ships effectively. The comparative results analysis demonstrates the betterment of the OMRCNN-SHD technique over the current methods with the maximum accuracy of 98.63%. [Display omitted] [ABSTRACT FROM AUTHOR]

Concrete privacy endeavours to give Confidentiality, Integrity, and Availability (CIA) measures to secure traffic streams in sensitive healthcare applications are a necessity. When talking about the access of sensitive healthcare data or the confidentiality of highly esteemed data, the first thing that comes to mind is that it should be secured. Sensitive healthcare data needs to be protected to restrict illegal access, exposure, and/or manipulation. As there is no such protection by which we can make our systems fully secure, the most acceptable methodology is to perform layered modelling of safety measures. This paper intends to provide a mathematical description of the Modular Encryption Standard (MES) and the augmentation of condition-centric risk monitoring of confidential information to provide a layered model for securing healthcare data confidentiality. Decision-making regarding the risk monitoring strategy of MES is augmented using a machine learning approach based on a Fuzzy Inference System amalgamated with Neural Networks. Result analysis shows that MES has less than a 0.005 error-rate and a 97% precision-rate, which elucidates its desideratum towards increasing security risks. [Display omitted] • Propose an approach for securing healthcare data confidentiality. • Provide mathematical modelling for the entire working of Modular Encryption Standard. • Enhance the decision-making strategy of Modular Encryption Standard using fuzzy logic. [ABSTRACT FROM AUTHOR]

• Categorized the attacks through the deep-learning method employing existing datasets. • Proposed a framework to add IDS as a program within IoT networks. • Designed a protection strategy for IoT network to maintain its integrity and provide its availability to legitimate users seamlessly. The Internet of Things (IoT) idea has been developed to enhance people's lives by delivering a diverse range of smart interconnected devices and applications in several domains. However, security threats are main critical challenges for the devices in an IoT environment. Many approaches have been proposed to secure IoT appliances in state of the art, still advancement is desirable. Machine learning has demonstrated a capability to detect patterns when other methodologies have collapsed. One advanced method to enhance IoT security is to employ deep learning. This formulates a seamless option for anomaly-based detection. This paper presents a CNN-based approach for anomaly-based intrusion detection systems (IDS) that takes advantage of IoT's power, providing qualities to efficiently examine whole traffic across the IoT. The proposed model shows ability to detect any possible intrusion and abnormal traffic behavior. The model is trained and tested using the NID Dataset and BoT-IoT datasets and achieved an accuracy of 99.51% and 92.85%, respectively. [Display omitted] [ABSTRACT FROM AUTHOR]

Deep neural networks are the state of the art for various machine learning problems dealing with large amounts of rich sensor data. It is often desirable to evaluate these models on edge devices instead of relying on cloud computing. In this paper, we perform traffic counting using surveillance cameras. Edge computing is required as only aggregated counts should leave the device and not the privacy sensitive video frames. Unfortunately, only small object detection models are suited for edge devices which results in sub-optimal performance. We introduce location specific models that are each trained for one specific camera. The model does not need to generalize to other locations. We show that smaller specialized models can outperform large general purpose models. We propose an automated way to train these small models without human intervention. We experimentally show that we can achieve a similar counting accuracy with 5x fewer parameters than state-of-the-art techniques. • We obtain high traffic counting accuracy in the real world using lightweight models. • We introduce location specific models that process the data of a single camera. • Since the models only need to work well for a single camera, they can be very small. • The location specific models can be trained automatically without human intervention. [ABSTRACT FROM AUTHOR]

Today, a significant number of biomedical data is generated continuously from various biomedical equipment and experiments due to rapid technological improvements in medical sciences. Effective biomedical data analysis, such as extracting biological and diagnostically significant features, is a very challenging task. This paper proposes hybrid Machine Learning Classification Techniques based on ensemble technique with Enhanced-Grey Wolf Optimization (E-GWO) feature selection algorithm to analyze these complex biomedical data. We combined five biomedical heart disease data sets for the experimental work: Cleveland, Long-Beach-VA, Switzerland, Hungarians, and Statlog. New hybrid Machine Learning Classification Techniques classifiers like Naive Bayes Bagging Technique (NBBT), Random Forest Bagging Technique (RFBT), Decision Tree Bagging Technique (DTBT), K-Nearest Neighbors Bagging Technique (KNNBT), Neural Network Bagging Technique (NNBT), Gradient Boosting Boosting Technique (GBBT), and Adaptive Boosting Boosting Technique (ABBT) are developed with bagging and boosting methods. Accuracy, Recall, Precision, F1-score, Specificity, Error Rate, G-mean, False Negative Rate (FNR), False Positive Rate (FPR), and Negative Predictive Value (NPV) are used to evaluate hybrid techniques. The experimental result shows that the developed hybrid classifier RFBT achieves the highest accuracy of 99.26% with E-GWO feature selection. The proposed method has 11.90% improved the accuracy to the conventional model. [ABSTRACT FROM AUTHOR]

With the recent proliferation of cyber physical systems (CPSs), there is a growing demand for reliable anomaly detection systems. In this paper, we propose a new ensemble learning approach for anomaly detection that utilizes the extraction of specific features tailored to anomaly detection problems. Whereas typical principal component analysis (PCA) selects principal components (PCs) associated with high variances, our proposed method also leverages PCs with low variances to account for unexpressed variations in the training data. The extracted features are then fed into conventional learning models such as support vector machines or recurrent neural networks. Since each PC can be particularly good at detecting certain types of attacks, classifiers based on different combinations of selected PCs are further combined as an ensemble. Our results show that the ensemble approach improves the overall accuracy and helps detect diverse types of unknown attacks as well. Furthermore, our simple yet effective and flexible approach can easily be deployed to various CPS environments of increasing complexity. [Display omitted] • We provide a new way to utilize PCA in ensemble frameworks for anomaly detection. • Classifiers are learned using different combinations of high- and low-variance PCs. • The ensemble classifier can detect various types of unseen anomaly attacks. • Our framework can be easily equipped with other machine learning detection models. [ABSTRACT FROM AUTHOR]

Software-Defined Networking (SDN) is a programmable network architecture that allows network devices to be controlled remotely, but it is still highly susceptible to traditional attacks such as Address Resolution Protocol (ARP) Poisoning, ARP Flooding, and others. The classification of benign network traffic from ARP Poison and ARP Flooding attacks is presented in this paper employing machine learning (ML) techniques. A python application is developed at the SDN controller using Mininet that collects and logs the features required to detect the attack into a file known as a traffic dataset. This dataset is used to train the ML model and detect the attacks. The hybrid model of Convolution Neural Network-Long Short Term Memory (CNN-LSTM) model out-performs the other ML models with an accuracy score of 99.73%. During the attack, a high CPU utilization of more than 97% and a high memory usage serve as experimental evidence. The attack detection time of 63000 microseconds also demonstrates the efficiency of attack detection. [Display omitted] • The author proposed an SDN-specific ARP attack dataset that can be used by the research community. • Using various features and promising deep learning algorithms are employed in order to detect the ARP Poison and Flood attacks. • Attack traffic is distinguished from benign traffic by employing various deep learning algorithms trained on the aforementioned SDN Dataset. [ABSTRACT FROM AUTHOR]

Message passing is a class of extremely powerful distributed iterative algorithms based on probabilistic graphic model, in which little computation performed per iteration on minimal data structure. As a prototypical message-passing algorithm, belief propagation (BP) algorithm has wide applications in various fields of coding theory, machine learning and combinatorial optimization. Due to its distributed and iterative nature, BP algorithm can run effectively and fast on large data networks. In this paper, we study the behavior of Min-Sum BP algorithm for the capacitated directed Chinese postman problem (CDCP). We derive the iterative process of message passing for solving CDCP. As the main result, for any weighted digraph G of size n , if the weight on each edge is nonnegative integral, then our algorithm converges to the optimal solution of CDCP after O (w ∗ n 2) iterations, provided that CDCP has a unique optimal solution, where w ∗ = max { w e : e ∈ E (G) }. [Display omitted] • Message passing is a class of extremely powerful distributed iterative algorithms based on probabilistic graphic model. • Due to its distributed and iterative nature, belief propagation (BP) algorithm can run effectively and fast on large data networks. • BP-based algorithms can also run fast on a large data network in synchronous circumstances. • Min-Sum BP algorithm could converge to the optimal solution of the capacitated directed Chinese postman problem after O (w m a x n 2) iterations. [ABSTRACT FROM AUTHOR]

In this paper we propose a Timing Recovery Loop for PSK and QAM modulations based on swarm Reinforcement Learning, suitable for FPGA implementation. We apply the Q-RTS algorithm, a hardware-oriented multi-agent version of Q-Learning, to a symbol synchronizer. One agent is in charge to synchronize the In-phase component and a second agent is applied to the Quadrature component. If compared to a loop based on a single-agent Q-Learning, we obtain improved synchronization capabilities in terms of recovery time and immunity to sub-optimal sampling. The Q-RTS timing recovery is up to 3 times faster than its Q-Learning counterpart. The implementation results show a low power consumption and a high throughput allowing the proposed synchronizer to be used in high-speed telecommunications systems. [Display omitted] • Telecommunications is a new field where Reinforcement Learning is applied. • Synchronization of PSK-QAM symbols is achieved via a multi-agent approach. • Multi-agent algorithms show an increased convergence speed instead of single ones. [ABSTRACT FROM AUTHOR]

By combining the Internet of Things (IoT) and Artificial Intelligence (AI), new augmentations and enhancements are realized, resulting in environment-aware systems that can enable intelligent decision making, with one such example being smart satellite networks. However, smart satellite networks attract the attention of hackers, resulting in them being targeted by severe cyber-attacks that can compromise their integrity, affect their availability, or breach the confidentiality of the data they generate, collect or rout. Lately, several prominent cyber-attacking scenarios have been observed to target IoT-enabled systems, resulting in loss, alteration, or exfiltration of data, disabling of devices, (distributed) denial of services, the formation of botnets, and lateral movement into otherwise secured networks. In this paper, we propose a network forensic framework based on Deep Learning (DL), the so-called Intelligent Satellite Deep Learning Network Forensic (INSAT-DLNF), for the detection and tracing of cyber-attack activities targeting smart satellite networks. For this framework, we trained a Long Short-term Memory Recurrent Neural Network (LSTM-RNN) and Gated Recurrent Unit (GRU) and compared their performances to five supervised and two unsupervised Machine Learning (ML) algorithms. The experimental results indicate that ML and DL algorithms can be employed effectively for the discovery and tracing of cyber-attacks, resulting in more adaptive and resilient cyber-security solutions, compared to legacy forensic tools that cannot discover zero-day attack surfaces and vectors. [Display omitted] • Smart Satellites vulnerable to IoT-centric cyber attacks. • Intelligent deep learning network forensic framework for smart satellite networks. • Big data analysis, for investigating attack events in smart satellite IoT networks. • Assess and compare network forensic methods using deep/machine learning. [ABSTRACT FROM AUTHOR]

Effective and timely fault detection and status monitoring of the industrial production process is essential to fully guarantee the operational safety. However, massive multi-source heterogeneous data analysis is facing many challenges. This paper proposes a process monitoring model combined with a parallel deep learning algorithm and Principal Component Analysis (PCA) method. Firstly, PCA is applied to realize fault diagnosis and extract fault characteristic variables, thus abnormal conditions can be revealed. In order to reduce the complexity of processing massive data, a parallel deep learning framework consisting of multi-models of convolutional neural network (CNN) and Long-Short Term Memory (LSTM) neural network is proposed to effectively predict the target variable status. Comprehensive experiments are taken under two real-system scenarios, and comparisons are made against four traditional neural network models to demonstrate its practicability and effectiveness. The generated results clearly show that the CNN-LSTM parallel model combined with PCA outperform other popular models due to its merged advantages of accurate time series prediction and effective fault feature extraction. [Display omitted] • A parallel deep learning approach with feature extraction and long time series prediction. • Multi-source unsupervised faulty variable extraction and detection can be realized. • Successfully solve the distribution power grid fault diagnosis in multiple locations. [ABSTRACT FROM AUTHOR]

Healthcare applications demand systematic approaches to eradicate inevitable human errors to design a framework that systematically eliminates cyber-threats. The key focus of this paper is to provide a comprehensive survey on the use of modern enabling technologies, such as the Internet of Things (IoT), 5G networks, artificial intelligence (AI), and big data analytics, for providing secure and resilient healthcare solutions. A detailed taxonomy of existing technologies has been demonstrated for tackling various healthcare problems, along with their security-related issues in handling healthcare data. The application areas of each of the emerging technologies, along with their security aspects, are explained. Furthermore, an IoT-enabled smart pill bottle prototype is designed and illustrated as a case study for providing better understanding of the subject. Finally, various key research challenges are summarized with future research directions. [Display omitted] • The survey provides a deep discussion on the use of: IoT-based secure healthcare solutions for provisioning prominent healthcare applications. • 5G services and their secure frameworks for healthcare applications. • AI for security enhancement in healthcare systems along with the specific application areas. • Big Data analytics for ensuring trustworthy healthcare systems. • Emerging technologies for the design and development of smart and secure IoT-pill bottle (Case Study). [ABSTRACT FROM AUTHOR]

IoT (Internet of Things) systems are still facing a great number of attacks due to their integration in several areas of life. The most-reported attacks against IoT systems are "Denial of Service" (DoS) and "Distributed Denial of Service" (DDoS) attacks. In this paper, we investigate DoS/DDoS attacks detection for IoT using machine learning techniques. We propose a new architecture composed of two components: DoS/DDoS detection and DoS/DDoS mitigation. The detection component provides fine-granularity detection, as it identifies the specific type of attack, and the packet type used in the attack. In this way, it is possible to apply the corresponding mitigation countermeasure on specific packet types. The proposed DoS/DDoS detection component is a multi-class classifier that adopts the "Looking-Back" concept, and is evaluated on the Bot-IoT dataset. Evaluation results show promising results as a Looking-Back-enabled Random Forest classifier achieves an accuracy of 99.81%. [ABSTRACT FROM AUTHOR]

In recent years, the residential load forecasting problem has been gaining renewed interest due to the advent of Smart Meters and Data Analytics. A novel hybrid method based on Empirical Mode Decomposition (EMD) in tandem with Extreme Learning Machine (ELM) is proposed in this paper to improve the forecast accuracy of residential load signals derived from Smart Meter data. Three state-of-the-art machine learning methods, namely Artificial Neural Network (ANN), Support Vector Regression (SVR), and ELM, are selected for performance comparison. It is observed from the results that the proposed method is found effective in picking the peaks that are usually present in residential loads and hence improved the forecast accuracy. Further, the results show that the performance of EMD based models is improved when the test data is characterized by more peaks. Smart*, a public dataset containing residential load measurements, is used for evaluation. [Display omitted] • A hybrid method is proposed for accurate prediction of Residential Loads. • Proposed method is validated using a real-time residential smart meter data. • Three hybrid and three non-hybrid models are evaluated. • Proposed method improves the prediction accuracy when more peaks are present. • Hybrid method based on Extreme Learning Machines offer better performance. [ABSTRACT FROM AUTHOR]

In this paper, we develop a feature extraction model using the amalgamation of machine and deep learning techniques for Industrial Internet of Thing (IIoT). We train the model with the most effective feature set evaluated using machine learning algorithms and deep learning feature extraction methods. We test these features with deep learning based network models for validation. We consider error metrics and accuracy as the major factors for combining the machine learning and deep learning techniques. The error rates are analysed using mean square error which show low error rate for the subset than the model tested for full dataset. Further, mean square error, accuracy rate and false values are analysed to test the performance of the proposed model. The comparative analysis with IIoT dataset and existing methods show that our approach is 25% better than the others. [Display omitted] • We address the multi-dimensional problem of the IIoT features to make an adequate feature extraction method. • We use machine learning classifier and wrapper subset evaluation technique with Naive-Bayes and meta bagging methods for feature selection. Deep learning auto encoders are implemented as a part of dimensionality reduction techniques and provides a amalgamation for the appropriateness of the features. We verify them with Principal Component Analysis (PCA) and ranking. Thus, only relevant features are extracted for analysis. • We create network models for analysing detection using deep learning Feed forward neural network and Elman Back propagation techniques for classification and detection with various training methods. [ABSTRACT FROM AUTHOR]

Fraud risks and threats are the critical concerns in Banking Cyber–Physical System (BCPS). Alongside the other banking frauds (i.e., forgery, altered cheques, identity theft, and impersonation), suspicious transaction detection is crucial in fraud detection. It is generally unfeasible to perform early scammer detection. Alleviating BCPS threats requires a blend of risk infiltration, sensing, and detection. A secure and robust approach is considerably required for tracking the interacting agent's provided credentials. The classical approaches provided in the literature are less competent, reliable, and accurate. Accordingly, in this paper, we provide the fastest, reliable, and efficient fraud detection approach. We use cognitive computing and quantum computing-based suspicious entity detection in the BCPS for the post-quantum era. The proposed approach achieves 97.04% of precision and 0.03% of error-rate against categorizing transactions as genuine or fraudulent. [Display omitted] • Propose a reliable, intelligent scheme for suspicious transaction detection in a BCPS. • Performs cognitive computing and quantum computing-based suspicious transaction detection. • Proposed approach efficiently assists in proper and timely secure responding. [ABSTRACT FROM AUTHOR]

With the rapid growth of threats and diversity in the manner of attack, Internet of things (IoT) systems has major challenges in providing methods to detect security vulnerabilities and attacks. There have been increasing developments of many detection tools and methods using full-time series data during malware execution based on machine learning/deep learning. However, the effectiveness of existing works is tightly bound by the requirement to use full-time series data. On the other hand, an earlier detection would help propose better solutions to respond to the IoT Botnet. Therefore, it mitigating the damage from potential attacks. In this paper, going beyond the full-time series data-based methods, we propose a collaborative machine learning model to effectively automate the early detection of IoT Botnet based on many features. The proposed model is 99.37% accurate on a dataset of 5023 IoT botnet and 3888 benign samples. [Display omitted] • IoT Botnet could be early detected by integrating multiple data sources. • Performance of proposed method not significantly increases over 180 s runtime and 300 different system-calls. • Basic machine learning model such as SVM, Random Forest... give promising results. [ABSTRACT FROM AUTHOR]

• The combination of mobile devices and edge computing reduces the total computing overhead and makes resource allocation reasonable. • Evaluate system performance by the weighted sum of delay and energy consumption. • Change the parameter update method to accelerate the convergence speed of the deep reinforcement learning algorithm. • A neural network is used to approximate the behavior value function, and a target q network is used to update the target. With the arrival of the 5th generation mobile networks (5 G) era, the data needed by mobile devices (MDs) is explosively growing. High-consumption, low-latency applications are huge challenges for resource-constrained Internet of things (IoT) devices. Mobile edge computing overcomes the limitations of computing resources on MDs by offloading tasks generated by MDs and assigning them to nearby MEC servers. Therefore, mobile edge computing (MEC) becomes important. This paper presents a task offloading strategy for the multi-device multi-server system. To meet the task requirements of different MDs, we formulate an overhead minimization problem to optimize the delay and energy consumption of the system. We propose the Double Deep Q Network (Double-DQN) algorithm to perform location selection strategies for tasks generated on the mobile devices and allocate respective computing resources. Simulation results show that the algorithm can allocate resources reasonably and reduce the overhead of the entire system. [Display omitted] [ABSTRACT FROM AUTHOR]

• Attack detection of smart home based on Internet of things security architecture • The two-layer data preprocessing method suitable for massive data scene of smart home • Detection model composed of binary classifiers improving the accuracy of malicious behavior With the development of Internet of Things and the increasingly rampant malicious network activities, higher requirements are put forward for security to detect malicious behavior and prevent attackers from obtaining sensitive data in the smart home environment. In this paper, an intrusion detection system is proposed to detect and classify abnormal behavior in the smart home environment. The two-layer feature processing method based on random forest and principal component analysis can reduce the loss of data information and is suitable for massive data. The three-layer detection model can detect four common attacks with binary classifiers and effectively improve the accuracy. The experimental evaluation of the proposed model is conducted using the real smart home traffic dataset and achieves a classification accuracy of 95.90%. The experimental results show that our model has a good performance in detecting and classifying malicious attacks in the smart home. [ABSTRACT FROM AUTHOR]

● Design a privacy preservation strategy for a secure edge intelligence. ● Federated machine learning mechanism for privacy-enhanced edge intelligence model. ● Beyond 5G networks with Paillier Homomorphic Encryption and differential privacy. ● Artificial Immune Intrusion Detection System to monitor and classify the nodes. ● Edge intelligence solutions with data-driven machine learning algorithms and cybersecurity. Beyond 5G networks integrating 5G technology offer efficient services globally with sustainable higher capacity and much lower latency across various applications. Moving into the beyond 5G networks, edge intelligence solutions with data-driven machine learning algorithms and cybersecurity paradigms have become crucial among various real-time applications, including smart transportation, smart health, etc. Since the data gets transferred continuously from the edge devices to the dedicated computing in any edge computing environment, it experiences a higher risk of vulnerability and complexity measures. In this view, this paper firstly defines a federated machine learning mechanism for privacy-enhanced edge intelligence model Beyond 5G networks with Paillier Homomorphic Encryption and differential privacy. Secondly, an Artificial Immune Intrusion Detection System has been designed to monitor and classify the nodes resulting in an anomaly in the edge network so that the network can experience smooth and secure data transmission as per the requirement. The experiments and comparison results show that the proposed system is more optimal and secure than the existing edge security models. [Display omitted] [ABSTRACT FROM AUTHOR]

• A cross-layer parallel attention network is proposed to further refine the semantic information of the high-level fusion features and the fine-grained information of the low-level fusion features. • A multiscale perception module is proposed to improve the robustness of the network to object scale change. • An effective anchor box allocation strategy is proposed to detect small objects at the low-level feature maps. • Our model can achieve mAP of 88.1% on the GDUT-HWD dataset and mAP of 80.5% on the PASCAL VOC 2007 dataset respectively. To solve the problem of low accuracy in existing safety helmet detection methods, a novel object detection algorithm based on Single Shot Multibox Detector (SSD) is proposed in this paper. The algorithm uses the spatial attention mechanism for low-level features and the channel attention mechanism for high-level features, this cross-layer attention mechanism can further refine the feature information of the object region. The proposed detection algorithm introduces a feature pyramid and multiscale perception module to improve its robustness to object scale change. In addition, an effective anchor box adaptive adjustment method is designed to adaptively adjust the scale distribution of each layer of the anchor boxes based on the feature map size. Experiment results demonstrate that our detection model has mean Average Precision (mAP) of 88.1% and 80.5% on helmet dataset and VOC 2007 dataset respectively, which is better than baseline by 15.65% and 3.4%. The network framework of CASP detection algorithm for safety helmet-wearing detection. [Display omitted] [ABSTRACT FROM AUTHOR]

Coronavirus is an infectious life-threatening disease and is mainly transmitted through infected person coughs, sneezes, or exhales. This disease is a global challenge that demands advanced solutions to address multiple dimensions of this pandemic for health and wellbeing. Different types of medical and technological-based solutions have been proposed to control and treat COVID-19. Machine learning is one of the technologies used in Magnetic Resonance Imaging (MRI) classification whereas nature-inspired algorithms are also adopted for image optimization. In this paper, we combined the machine learning and nature-inspired algorithm for brain MRI images of COVID-19 patients namely Machine Learning and Nature Inspired Model for Coronavirus (MLNI-COVID-19). This model improves the MRI image classification and optimization for better diagnosis. This model will improve the overall performance especially the area of brain images that is neglected due to the unavailability of the dataset. COVID-19 has a serious impact on the patient brain. The proposed model will help to improve the diagnosis process for better medical decisions and performance. The proposed model is evaluated with existing algorithms and achieved better performance in terms of sensitivity, specificity, and accuracy. [ABSTRACT FROM AUTHOR]

Non-technical losses stand for the energy consumed but not billed, affecting the energy grid as a whole. Such an issue somehow prevails in developing countries, harming the quality of energy and preventing social programs benefit from tax revenues. Machine learning techniques can help mitigate it by mining information from fraudsters and legal users for further decision-making. In this paper, we deal with a steady increase of dataset size, i.e., the incremental learning problem, which can cope with datasets regularly provided by energy companies, requiring the learner to be updated constantly. Since repeating the entire learning process might be prohibitive, adjusting the model to the new data shows to be a better choice. We propose an incremental Optimum-Path Forest approach with k -nn neighborhood that is considerably more efficient for training than its counterpart version, with experiments validated in general-purpose datasets and also in the context of non-technical losses identification. [Display omitted] • A new incremental learning algorithm for the Optimum-Path Forest classifier. • Proposed approach outperformed its counterpart in terms of efficiency. • First OPF-based incremental learning applied to non-technical losses identification. [ABSTRACT FROM AUTHOR]

• Electricity fraud detection in conventional meters. • Combining ML: supervised and unsupervised algorithms. • Extensive feature engineering. • Multivariate Gaussian distribution for anomaly detection. • Light Gradient Boosting algorithm: performance and tuning. Although fraud in electricity consumption is easier to detect when consumption is recorded hourly by smart meters, in most developing countries, where the propensity for fraud is higher, conventional meters are not yet affordable. Fraud detection is easier with time series data-logging due to the periodicity and variability of consumption that reveals deviations from a regular consumption pattern. In contrast, fraud detection with conventional meters remains a significant challenge because anomalies in consumption are well hidden within the normal consumption of other consumers. In this paper, large datasets regarding consumers and invoice data from Tunisia are combined and investigated with several Machine Learning (ML) classification algorithms, to detect irregularities in electricity consumption. By performing extensive feature engineering, including multivariate Gaussian distribution, the efficiency of ensemble classifiers such as Light Gradient Boosting (LGB) outperforms other algorithms and achieves realistic performance from challenging, unbalanced and uncorrelated input datasets. [ABSTRACT FROM AUTHOR]

A new paradigm of IoT monitoring using sonar sensors and microphones is studied as a contactless alternative to the traditional devices for sleep disorders in this paper. A pair of sonar sensors are used to measure the distance from the bottom and side of the body to the bedside respectively, and the basic posture of the human body during sleep can be inferred by using machine learning. When a person sleeps still, two streams of sonar signals from two adjacent sides remain unchanged. Any movement would disrupt the stationary sonar streams. Hence it could be detected as change of posture from lying still. Different sonar patterns could be recognized as specific sleeping postures by some non-linear machine learning model. This simple and novel solution could potentially be used as an alternative or supplementary to video analytics which can feedback to the user about their sleeping pattern. A new data transformation method namely Kennard-stone Balance (KSB) algorithm is also proposed for simplifying the data streams and enhancing the accuracy of the machine learning model. Simulation results show the feasibility of this sonar method, and KSB is able to improve the pattern recognition performance. [ABSTRACT FROM AUTHOR]

We live in an interconnected and pervasive world where huge amount of data are collected every second. Fully exploiting data through advanced analytics, machine learning and artificial intelligence, becomes crucial for businesses, from micro to large enterprises, resulting in a key advantage (or shortcoming) in the global market competition, as well as in a strong market driver for business analytics solutions. This scenario is deeply changing the security landscape, introducing new risks and threats that affect security and privacy of systems, on one side, and safety of users, on the other side. Many domains that can benefit from novel solutions based on data analytics have stringent security requirements to fulfill. The Energy domain's Smart Grid is a major example of systems at the crossroads of security and data-driven intelligence. The Smart Grid plays a crucial role in modern energy infrastructure. However, it must face two major challenges related to security: managing front-end intelligent devices such as power assets and smart meters securely, and protecting the huge amount of data received from these devices. Starting from these considerations, setting up proper analytics is a complex problem because security controls could have the undesired side effect of decreasing the accuracy of the analytics themselves. This is even more critical when the configuration of security controls is let to the security expert, who often has only basic skills in data science. In this paper, we propose a solution based on the concept of Model-Based Big Data Analytics-as-a-Service (MBDAaaS) that bridges the gap between security experts and data scientists. Our solution acts as a middleware allowing a security expert and a data scientist to collaborate to the deployment of an analytics addressing their needs. [Display omitted] • Big Data Analytics-as-a-Service bridges the gap between security experts and data scientists. • Security experts and data scientists can seamlessly collaborate to the deployment of an analytics process for cybersecurity. • Big Data Analytics-as-a-Service reduces the impact of the lack of professionals with both security and data science expertise. [ABSTRACT FROM AUTHOR]

• Screening of crystal particles through machine learning methods. • Crystal detection method based on Gabor filter and SVM. • The crystal detection method based on Gabor filter and SVM plays an important role in the detection of grain defects. During the slicing process, thermoelectric cooler (TEC) component are prone to cause various defects, such as cracks, pockmarks, and other defects of various shapes and sizes. The presence of these surface defects will lead to a decline in product performance. At present, the method of screening TEC component is mainly manual detection. This method is affected by the strong subjectivity of human beings and the limited spatial resolution. Therefore, a method based on machine learning is proposed to overcome the above shortcomings and obtain product accuracy that is difficult for artificial vision to achieve We use Gabor filters to extract texture information in the TEC component image and use it as a classification feature. At the same time, principal component analysis (PCA) is used to select the classification features. For the number of defect types, we use support vector machine (SVM) to classify unknown defects. By extracting the features of the TEC component image and using it for SVM training. Finally, the unknown defect image is used as input to obtain the defect category By testing the TEC simultaneously with other machine learning methods in this paper, the superiority of this method is proved. At the same time, it also overcomes a series of problems such as slow manual detection speed, low reliability, poor consistency and stability of test results, product quality cannot be improved, and low detection efficiency Based on our experimental analysis, the proposed method can effectively improve the classification effect of TEC component defects. Compared with other classifiers, the effectiveness of this method is further verified. [ABSTRACT FROM AUTHOR]

With improvements in human living standard, people's demands on food quality are getting higher and higher. Effective food recognition algorithms are needed to obtain more useful food information. In order to solve the problem of low accuracy and slow speed of food recognition algorithms, a new food recognition algorithm based on CNN algorithm is proposed. First, the proposed algorithm preprocess the food images which are collected from the internet. And then use the traditional convolution extract the features from food images. The jumping convolution which is designed in this paper to extract food features jumping and combines the features from traditional convolutions. This algorithm cannot only solve the food recognition problem effectively, but also reduce the calculation parameters. Compared with the experimental results of other deep learning networks, the proposed algorithm has a good effect, and can recognize the food quickly and reduce the training time. [ABSTRACT FROM AUTHOR]

In this era of modernization, digital technology plays a major role in all facets of life. We are accustomed to using computers and smartphones to access information, create, express, communicate, and collaborate, contributing to our personal, social, and professional survival at a comfortable level. As the legitimate use of digital space has grown, so have the prospects for malice via swindlers, blackmailers, vandals, and other criminals, to benefit by creating and propagating malware. While the main motive of malware designers is to make surplus profit illegally, other trivial motives range from activism and pranks, espionage, cyber theft, as well as serious crimes like privacy breaches. Policing the expanding attack surface needs the development of competent anti-malware solutions with more generalization to detect even zero-day malware and resolve their incidences with minimal human intervention. Anti-malware solutions pick up effective strategies from static, dynamic, reverse engineering, and advanced machine learning techniques. An arms-race situation always exists between the malware authors and the anti-malware community. In this paper, we present a comprehensive investigation of computer-based and mobile-based malware, their countermeasures, and various detection methods. We also discuss the additional issues and the challenges of malware detection and finally highlight a few open research issues, directing the trends of malware/anti-malware development. [ABSTRACT FROM AUTHOR]

In this research, some of the issues that arise from the scalarization of the multi-objective optimization problem in the Advantage Actor–Critic (A2C) reinforcement learning algorithm are investigated. The paper shows how a naive scalarization can lead to gradients overlapping. Furthermore, the possibility that the entropy regularization term can be a source of uncontrolled noise is discussed. With respect to the above issues, a technique to avoid gradient overlapping is proposed, while keeping the same loss formulation. Moreover, a method to avoid the uncontrolled noise, by sampling the actions from distributions with a desired minimum entropy, is investigated. Pilot experiments have been carried out to show how the proposed method speeds up the training. The proposed approach can be applied to any Advantage-based Reinforcement Learning algorithm. [Display omitted] • Issues of the Advantage Actor–Critic (A2C) gradient-based optimization are addressed. • The proposed variant of the A2C avoids gradient overlapping and controls the entropy. • The speedup with respect to A2C is better for increasing environment complexity. [ABSTRACT FROM AUTHOR]

• NIDS is an essential part of the network security. • Deep learning is one of the field of machine learning. • We proposed deep learning based feature extraction in IDS. • This method can optimize pattern matching engine in intrusion detection. • We performed a case study on the pattern matching engine. Deep learning has gained popularity for feature extraction in the field of Network Intrusion Detection and Prevention System (NIDPS) to extract the patterns matching and secure the networks by detecting the unknown and malicious activities. The malicious activities and security attacks are disturbing the normal operations of networks. The new attacks are difficult to monitor due to their new features and pattern types. Different types of methods have been adopted for feature extraction and pattern matching. Deep learning is one of them and subfield of machine learning where it solves the optimization issues layer-wise by looking at the deep structure. On the other hand, pattern matching is another considerable method for intrusion detection due to its variety of applications. However, pattern matching methods are consuming more than 70% of the total running time and cause overhead. In this paper, we propose two methods including Deep Learning-based Feature Extraction (DLFE) and Optimization of Pattern Matching (OPM) for NIDPS systems to optimizes the pattern matching engine in intrusion detection. The experiments are performed by using the snort ruleset for pattern matching and obtained the results. The experiment results show the better performance of proposed methods in terms of time, throughput, and memory. [ABSTRACT FROM AUTHOR]

• Trojan malware detection in the cloud-computing infrastructure. • Using semantic algorithms to detect malware, in particular unknown ones. • Ability to detect malicious mobile malware with very high accuracy. • Using subgraphs semantic homeomorphism coefficient innovation to classify Trojans. • Combining DNN GMDH algorithm and semantic methods to detect malware automatically. There are several approaches for preventing mobile devices from malware intrusion, but most of them suffer from the insufficient accuracy required for detecting Trojan malware. A combination of semantic and machine learning techniques can be effective in preventing intrusions. In this paper, we have used a hierarchical semantic approach to convert numerical and string data to meaningful values, Subgraph Semantic Homomorphism Coefficient (SSHC) to select optimal features, and Group Method of Data Handling (GMDH) deep neural network (DNN) algorithm to detect malware via a cloud-computing infrastructure. To evaluate our model, Android Trojan Dataset has been used. After evaluation, the accuracy reached 99.91%, which was improved by about 5.25% compared to StormDroid, Drebin, and KuafuDet models. Also, the accuracy was improved by about 10.4% and 31.9% compared to machine learning based approaches of Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbor (KNN), in the state-of-the-art KuafuDet model, respectively. [Display omitted] [ABSTRACT FROM AUTHOR]

Cybersecurity in the automotive sector is becoming increasingly important as modern vehicles can be exposed to cybersecurity threats and regulatory fraud. Sensor identification and authentication is an important function in the vehicle life cycle. This paper proposes an alternative or complementary identification and authentication approach to cryptographic techniques based on the physical properties of the Hall sensor, which is commonly used in odometers and motion sensors. The application of different signal processing algorithms to an experimental dataset of 12 Hall sensors was evaluated. In particular, a dimensionality reduction with decimation filters was applied to improve the time efficiency of the identification and authentication process while achieving high identification accuracy (accuracy=95%) and high authentication accuracy. In addition, the results show that different transformations with different hyperparameters are able to generate a wide range of challenges and responses to support the authentication process. [ABSTRACT FROM AUTHOR]

Machine learning (ML) has demonstrated great potential to revolutionize the networking field. In this paper, we present a large-scale empirical study to evaluate the effectiveness of state-of-the-art ML algorithms for network application security. In our experiments, six classical ML algorithms and three neural network algorithms are evaluated over three networking datasets, KDDCup 99, NSL-KDD, and ADFA IDS 2017. Measurements are made between the non-optimized and optimized versions of ML algorithms. Furthermore, various training and testing ratios are experimented to assess each algorithm's optimal performance. The results revealed that optimizing ML algorithms could help achieve better performance in detecting networking attacks. In particular, the Decision Tree proved to be the most accurate and fastest algorithm in the classical ML while the Recurrent Neural Network achieved the best performance among neural network algorithms. [Display omitted] • Providing a comprehensive empirical evaluation of ML algorithms. • Applying optimization methods to evaluate performance with respect to networking attacks. • Providing accuracy results with and without optimization with different ML algorithms. [ABSTRACT FROM AUTHOR]

The rapid growth of network-related services in the last decade has produced a huge amount of sensitive data on the internet. But networks are very much prone to intrusions where unauthorized users attempt to access sensitive information and even disrupt the system. Building a competent network intrusion detection system (IDS) is necessary to prevent such attacks. IDSs generally use machine learning algorithms for classifying the attacks. But the features used for classification are not always suitable or sufficient. Besides, the number of intrusions is much less than the number of non-intrusions. Hence naive approaches may fail to provide acceptable performance due to this class imbalance. To counter this problem, in this paper, we propose a model that extracts useful features from the given features and then uses a deep learning algorithm to classify the intrusions. It is to be noted that underlying data points cannot be thought of as sampled from the same distribution, rather from two different distributions - one generic to all network intrusions, and the other specific to the domain. Keeping this fact in mind, we propose a unique Generic-Specific autoencoder architecture where the generic one learns the features that are common across all forms of network intrusions, and the specific ones learn features that are pertaining only to that domain. The model has been evaluated on the CICIDS2017 dataset, which is the largest dataset of this type available online, and we have set new benchmark results on this dataset. Source code of this work is available at: https://github.com/SoumyadeepThakur/Intrusion-AE [ABSTRACT FROM AUTHOR]

• Online learning with the increased classes is considered. • Two different types of the increased class are indicated. • The structure of OS-ELM is changed to adapt two different increased class types. • Results indicate that the proposed algorithm can adapt the increased class well. In this paper, we investigate the online learning problem in a specific open environment, i.e., the number of classes can be dynamically increased. Online sequential extreme learning machine (OS-ELM) is extended to address the problem of the increased classes. Specifically, two different increased classes scenarios are considered. The first scenario is that the new classes, which haven't appeared in the previous instances, emerge in the new received data. The other scenario is that in data stream, an old class is split into several new subclasses due to some specific reasons. For the first kind of scenario, OS-ELM is inserted an alternative output node which can be extended whenever the new class instances are received. While for the second kind of scenario, we adopt a hierarchical structure to adapt the new split classes. We adopt the simple experiments to show the effectiveness and feasibility of the proposed models. [ABSTRACT FROM AUTHOR]

Instance selection is an important problem in medical data mining. It focuses on selecting representative data samples from a given training set, whereas unrepresentative (or noisy) data samples are filtered out. This reduces the size of the training set, which then requires less storage space. In addition, when the instance selection algorithm was carefully chosen, a reduction in the training set so that it contains less noisy data can usually make the classifiers perform better than the ones without considering instance selection. In the literature, many instance selection algorithms have been proposed. However, different algorithms tend to use different criteria to determine the noisy data, making it difficult to find the best algorithm for different domain datasets. In other words, some algorithms may perform better than the others for some specific domain datasets, but may perform worse than others over other domain datasets. Instead of developing a novel algorithm that performs better than most other algorithms, this paper introduces a divide-and-conquer based instance selection (DCIS) framework that aims to improve the performance of each specific instance selection algorithm per se. Two well-known algorithms, i.e., DROP3 and IB3, are used as the baseline, and various small and large scale medical datasets are used in the experiments. Our results show that when DROP3 and IB3 are used to perform instance selection based on the DCIS framework, there is an improvement in the performance of the k-NN and SVM classifiers over the ones by the DROP3 and IB3 baselines, respectively. [ABSTRACT FROM AUTHOR]