HSDDD: A Hybrid Scheme for the Detection of Distracted Driving through Fusion of Deep Learning and Handcrafted Features.

Traditional methods for behavior detection of distracted drivers are not capable of capturing driver behavior features related to complex temporal features. With the goal to improve transportation safety and to reduce fatal accidents on roads, this research article presents a Hybrid Scheme for the Detection of Distracted Driving called HSDDD. This scheme is based on a strategy of aggregating handcrafted and deep CNN features. HSDDD is based on three-tiered architecture. The three tiers are named as Coordination tier, Concatenation tier and Classification tier. We first obtain HOG features by using handcrafted algorithms, and then at the coordination tier, we leverage four deep CNN models including AlexNet, Inception V3, Resnet50 and VGG-16 for extracting DCNN features. DCNN extracted features are fused with HOG extracted features at the Concatenation tier. Then PCA is used as a feature selection technique. PCA takes both the extracted features and removes the redundant and irrelevant information, and it improves the classification performance. After feature fusion and feature selection, the two classifiers, KNN and SVM, at the Classification tier take the selected features and classify the ten classes of distracted driving behaviors. We evaluate our proposed scheme and observe its performance by using the accuracy metrics.

Avoiding Void Holes and Collisions with Reliable and Interference-Aware Routing in Underwater WSNs.

Sparse node deployment and dynamic network topology in underwater wireless sensor networks (UWSNs) result in void hole problem. In this paper, we present two interference-aware routing protocols for UWSNs (Intar: interference-aware routing; and Re-Intar: reliable and interference-aware routing). In proposed protocols, we use sender based approach to avoid the void hole. The beauty of the proposed schemes is that they not only avoid void hole but also reduce the probability of collision. The proposed Re-Intar also uses one-hop backward transmission at the source node to further improve the packet delivery ratio of the network. Simulation results verify the effectiveness of the proposed schemes in terms of end-to-end delay, packet delivery ratio and energy consumption.

Void Hole Avoidance for Reliable Data Delivery in IoT Enabled Underwater Wireless Sensor Networks.

Due to the limited availability of battery power of the acoustic node, an efficient utilization is desired. Additionally, the aquatic environment is harsh; therefore, the battery cannot be replaced, which leaves the network prone to sudden failures. Thus, an efficient node battery dissipation is required to prolong the network lifespan and optimize the available resources. In this paper, we propose four schemes: Adaptive transmission range in WDFAD-Depth-Based Routing (DBR) (A-DBR), Cluster-based WDFAD-DBR (C-DBR), Backward transmission-based WDFAD-DBR (B-DBR) and Collision Avoidance-based WDFAD-DBR (CA-DBR) for Internet of Things-enabled Underwater Wireless Sensor Networks (IoT, UWSNs). A-DBR adaptively adjusts its transmission range to avoid the void node for forwarding data packets at the sink, while C-DBR minimizes end-to-end delay along with energy consumption by making small clusters of nodes gather data. In continuous transmission range adjustment, energy consumption increases exponentially; thus, in B-DBR, a fall back recovery mechanism is used to find an alternative route to deliver the data packet at the destination node with minimal energy dissipation; whereas, CA-DBR uses a fall back mechanism along with the selection of the potential node that has the minimum number of neighbors to minimize collision on the acoustic channel. Simulation results show that our schemes outperform the baseline solution in terms of average packet delivery ratio, energy tax, end-to-end delay and accumulated propagation distance.

Neuro-XAI: Explainable deep learning framework based on deeplabV3+ and bayesian optimization for segmentation and classification of brain tumor in MRI scans.

The prevalence of brain tumor disorders is currently a global issue. In general, radiography, which includes a large number of images, is an efficient method for diagnosing these life-threatening disorders. The biggest issue in this area is that it takes a radiologist a long time and is physically strenuous to look at all the images. As a result, research into developing systems based on machine learning to assist radiologists in diagnosis continues to rise daily. Convolutional neural networks (CNNs), one type of deep learning approach, have been pivotal in achieving state-of-the-art results in several medical imaging applications, including the identification of brain tumors. CNN hyperparameters are typically set manually for segmentation and classification, which might take a while and increase the chance of using suboptimal hyperparameters for both tasks. Bayesian optimization is a useful method for updating the deep CNN's optimal hyperparameters. The CNN network, however, can be considered a "black box" model because of how difficult it is to comprehend the information it stores because of its complexity. Therefore, this problem can be solved by using Explainable Artificial Intelligence (XAI) tools, which provide doctors with a realistic explanation of CNN's assessments. Implementation of deep learning-based systems in real-time diagnosis is still rare. One of the causes could be that these methods don't quantify the Uncertainty in the predictions, which could undermine trust in the AI-based diagnosis of diseases. To be used in real-time medical diagnosis, CNN-based models must be realistic and appealing, and uncertainty needs to be evaluated. So, a novel three-phase strategy is proposed for segmenting and classifying brain tumors. Segmentation of brain tumors using the DeeplabV3+ model is first performed with tuning of hyperparameters using Bayesian optimization. For classification, features from state-of-the-art deep learning models Darknet53 and mobilenetv2 are extracted and fed to SVM for classification, and hyperparameters of SVM are also optimized using a Bayesian approach. The second step is to understand whatever portion of the images CNN uses for feature extraction using XAI algorithms. Using confusion entropy, the Uncertainty of the Bayesian optimized classifier is finally quantified. Based on a Bayesian-optimized deep learning framework, the experimental findings demonstrate that the proposed method outperforms earlier techniques, achieving a 97 % classification accuracy and a 0.98 global accuracy.

Extended Kalman Filter-Based Power Line Interference Canceller for Electrocardiogram Signal.

Cardiac diseases constitute a major root of global mortality and they are likely to persist. Electrocardiogram (ECG) is widely opted in clinics to detect countless heart illnesses. Numerous artifacts interfere with the ECG signal, and their elimination is vital to allow medical specialists to acquire valuable statistics from the ECG. The utmost artifact that is added to the ECG signal is power line interference (PLI). Numerous filtering methods have been employed in the literature to eliminate PLI from noisy ECG. This article proposes an extended Kalman filter (EKF)-based adaptive noise canceller (ANC) that comprises PLI frequency as a distinct model parameter. Thus, it is capable of tracking PLI with drifting frequency. The proposed canceller's performance is compared with state-space recursive least squares (SSRLSs) filter-based PLI canceling. The evaluation is carried out for four cases of PLI, that is, PLI with known amplitude and frequency, PLI with unknown amplitude and frequency, PLI with drifting amplitude and frequency, and PLI removal from a real-time ECG recording. The samples of the Massachusetts Institude of Technology (MIT)-Boston's Beth Israel Hospital (BIH) arrhythmia database are considered for the first three cases, whereas, for the fourth case, real ECG signal is taken from armed forces institude of cardiology, the national institude of heart diseases (AFIC/NIHD), Pakistan. Mean square error, frequency spectrum, and noise reduction are selected as performance metrics for comparison. Simulation results depict that the presented EKF-based ANC system outperforms the SSRLS-based ANC system and effectively eliminates PLI from ECG under all four investigated scenarios.

GestroNet: A Framework of Saliency Estimation and Optimal Deep Learning Features Based Gastrointestinal Diseases Detection and Classification.

In the last few years, artificial intelligence has shown a lot of promise in the medical domain for the diagnosis and classification of human infections. Several computerized techniques based on artificial intelligence (AI) have been introduced in the literature for gastrointestinal (GIT) diseases such as ulcer, bleeding, polyp, and a few others. Manual diagnosis of these infections is time consuming, expensive, and always requires an expert. As a result, computerized methods that can assist doctors as a second opinion in clinics are widely required. The key challenges of a computerized technique are accurate infected region segmentation because each infected region has a change of shape and location. Moreover, the inaccurate segmentation affects the accurate feature extraction that later impacts the classification accuracy. In this paper, we proposed an automated framework for GIT disease segmentation and classification based on deep saliency maps and Bayesian optimal deep learning feature selection. The proposed framework is made up of a few key steps, from preprocessing to classification. Original images are improved in the preprocessing step by employing a proposed contrast enhancement technique. In the following step, we proposed a deep saliency map for segmenting infected regions. The segmented regions are then used to train a pre-trained fine-tuned model called MobileNet-V2 using transfer learning. The fine-tuned model's hyperparameters were initialized using Bayesian optimization (BO). The average pooling layer is then used to extract features. However, several redundant features are discovered during the analysis phase and must be removed. As a result, we proposed a hybrid whale optimization algorithm for selecting the best features. Finally, the selected features are classified using an extreme learning machine classifier. The experiment was carried out on three datasets: Kvasir 1, Kvasir 2, and CUI Wah. The proposed framework achieved accuracy of 98.20, 98.02, and 99.61% on these three datasets, respectively. When compared to other methods, the proposed framework shows an improvement in accuracy.

Have You Been a Victim of COVID-19-Related Cyber Incidents? Survey, Taxonomy, and Mitigation Strategies.

Cybercriminals are constantly on the lookout for new attack vectors, and the recent COVID-19 pandemic is no exception. For example, social distancing measures have resulted in travel bans, lockdowns, and stay-at-home orders, consequently increasing the reliance on information and communications technologies, such as Zoom. Cybercriminals have also attempted to exploit the pandemic to facilitate a broad range of malicious activities, such as attempting to take over videoconferencing platforms used in online meetings/educational activities, information theft, and other fraudulent activities. This study briefly reviews some of the malicious cyber activities associated with COVID-19 and the potential mitigation solutions. We also propose an attack taxonomy, which (optimistically) will help guide future risk management and mitigation responses.

A New Random Walk for Replica Detection in WSNs.

Wireless Sensor Networks (WSNs) are vulnerable to Node Replication attacks or Clone attacks. Among all the existing clone detection protocols in WSNs, RAWL shows the most promising results by employing Simple Random Walk (SRW). More recently, RAND outperforms RAWL by incorporating Network Division with SRW. Both RAND and RAWL have used SRW for random selection of witness nodes which is problematic because of frequently revisiting the previously passed nodes that leads to longer delays, high expenditures of energy with lower probability that witness nodes intersect. To circumvent this problem, we propose to employ a new kind of constrained random walk, namely Single Stage Memory Random Walk and present a distributed technique called SSRWND (Single Stage Memory Random Walk with Network Division). In SSRWND, single stage memory random walk is combined with network division aiming to decrease the communication and memory costs while keeping the detection probability higher. Through intensive simulations it is verified that SSRWND guarantees higher witness node security with moderate communication and memory overheads. SSRWND is expedient for security oriented application fields of WSNs like military and medical.

Distributed clone detection in static wireless sensor networks: random walk with network division.

Wireless Sensor Networks (WSNs) are vulnerable to clone attacks or node replication attacks as they are deployed in hostile and unattended environments where they are deprived of physical protection, lacking physical tamper-resistance of sensor nodes. As a result, an adversary can easily capture and compromise sensor nodes and after replicating them, he inserts arbitrary number of clones/replicas into the network. If these clones are not efficiently detected, an adversary can be further capable to mount a wide variety of internal attacks which can emasculate the various protocols and sensor applications. Several solutions have been proposed in the literature to address the crucial problem of clone detection, which are not satisfactory as they suffer from some serious drawbacks. In this paper we propose a novel distributed solution called Random Walk with Network Division (RWND) for the detection of node replication attack in static WSNs which is based on claimer-reporter-witness framework and combines a simple random walk with network division. RWND detects clone(s) by following a claimer-reporter-witness framework and a random walk is employed within each area for the selection of witness nodes. Splitting the network into levels and areas makes clone detection more efficient and the high security of witness nodes is ensured with moderate communication and memory overheads. Our simulation results show that RWND outperforms the existing witness node based strategies with moderate communication and memory overheads.