Search

Countless cybercrime instances have shown the need for detecting and blocking obscene material from social media sites. Deep learning methods (DLMs) outperformed in recognizing obscene content flooded on many online platforms. However, these contemporary DLMs primarily treat the recognition of obscene content as a simple task of binary classification, rather than focusing on the labelling of obscene areas. Hence, many of these methods could not pay attention to the fact that misclassification samples are so diverse. Therefore, this paper focuses on two aspects (i) developing a deep learning model that could classify and label the obscene portion, and (ii) generating a labelled obscene image dataset with a wide variety of obscene samples to minimize the risks of inaccurate recognition. We have proposed a method named S3Pooling based bottleneck attention module (BAM) embedded MobileNetV2‐YOLOv3 (SBMYv3) for automatic detection of obscene content using an attention mechanism and a suitable pooling strategy. The key contributions of our article are: (i) generation of a well‐labelled obscene image dataset with a variety of augmentation strategies using Pix‐2‐Pix GAN (ii) modifications to the backend architecture of YOLOv3 using MobileNetV2 and BAM to ensure focused and accurate feature extraction, and (iii) selection of an optimal pooling strategy, that is, S3Pooling strategy, while taking the design of the feature extractor into account. The proposed SBMYv3 model outperformed other state‐of‐the‐art models with 99.26% testing accuracy, 99.39% recall, 99.13% precision, and 99.13% IoU values respectively. [ABSTRACT FROM AUTHOR]

Hand gestures are an effective method of communication, especially when we are communicating with people who cannot understand our spoken language. Furthermore, it is a key aspect to human–computer interaction. Understanding hand gestures is very important to ensure that listeners understand what speakers are attempting to communicate. Even though several researchers have proposed deep learning-based models for hand gesture recognition, the hyper-parameter tuning of these models is a relatively unexplored area. In this work, Convolutional Neural Networks (CNN) are used to classify hand gesture images. To tune the hyper-parameters of the CNN, a recently developed metaheuristic algorithm, namely, the Harris Hawks Optimization (HHO) algorithm, is used. Our in-depth comparative analysis proves that the proposed HHO-CNN hybrid model outperforms the existing models by attaining an Accuracy of 100%. [ABSTRACT FROM AUTHOR]

Breast cancer is one of the most widespread diseases in women worldwide. It leads to the second-largest mortality rate in women, especially in European countries. It occurs when malignant lumps that are cancerous start to grow in the breast cells. Accurate and early diagnosis can help in increasing survival rates against this disease. A computer-aided detection (CAD) system is necessary for radiologists to differentiate between normal and abnormal cell growth. This research consists of two parts; the first part involves a brief overview of the different image modalities, using a wide range of research databases to source information such as ultrasound, histography, and mammography to access various publications. The second part evaluates different machine learning techniques used to estimate breast cancer recurrence rates. The first step is to perform preprocessing, including eliminating missing values, data noise, and transformation. The dataset is divided as follows: 60% of the dataset is used for training, and the rest, 40%, is used for testing. We focus on minimizing type one false-positive rate (FPR) and type two false-negative rate (FNR) errors to improve accuracy and sensitivity. Our proposed model uses machine learning techniques such as support vector machine (SVM), logistic regression (LR), and K-nearest neighbor (KNN) to achieve better accuracy in breast cancer classification. Furthermore, we attain the highest accuracy of 97.7% with 0.01 FPR, 0.03 FNR, and an area under the ROC curve (AUC) score of 0.99. The results show that our proposed model successfully classifies breast tumors while overcoming previous research limitations. Finally, we summarize the paper with the future trends and challenges of the classification and segmentation in breast cancer detection. [ABSTRACT FROM AUTHOR]

Dry eye disease (DED) is a chronic eye disease and a common complication among the world's population. Evaporation of moisture from tear film or a decrease in tear production leads to an unstable tear film which causes DED. The tear film breakup time (TBUT) test is a common clinical test used to diagnose DED. In this test, DED is diagnosed by measuring the time at which the first breakup pattern appears on the tear film. TBUT test is subjective, labour-intensive and time-consuming. These weaknesses make a computer-aided diagnosis of DED highly desirable. The existing computer-aided DED detection techniques use expensive instruments for image acquisition which may not be available in all eye clinics. Moreover, among these techniques, TBUT-based DED detection techniques are limited to finding only tear film breakup area/time and do not identify the severity of DED, which can essentially be helpful to ophthalmologists in prescribing the right treatment. Additionally, a few challenges in developing a DED detection approach are less illuminated video, constant blinking of eyes in the videos, blurred video, and lack of public datasets. This paper presents a novel TBUT-based DED detection approach that detects the presence/absence of DED from TBUT video. In addition, the proposed approach accurately identifies the severity level of DED and further categorizes it as normal, moderate or severe based on the TBUT. The proposed approach exhibits high performance in classifying TBUT frames, detecting DED, and severity grading of TBUT video with an accuracy of 83%. Also, the correlation computed between the proposed approach and the Ophthalmologist's opinion is 90%, which reflects the noteworthy contribution of our proposed approach. [ABSTRACT FROM AUTHOR]

Predicting global solar activity is crucial for smart cities, especially for space activities, communication industries, and climate change monitoring. The recently developed models to predict solar activity based on stand-alone artificial intelligence, using machine and deep learning models, and hybrid models are promising. Yet they may not be effective at capturing simpler linear patterns in the data and often fail to provide reliable predictions due to their computational cost and complexity. This article proposed a novel residual hybrid machine learning method integrating linear regression machine learning, and deep learning neural networks for solving predictive accuracy in individual machine learning models that reduces complexity. The residual hybrid model leverages the capacities of the support vector machine (SVM) and long short-term memory neural network (LSTM) for hybrid SVM-LSTM model. The performance of the model is evaluated using the correlation coefficient, determination coefficient, root-mean-squared error (RMSE) and mean-absolute error. The simulation results indicated that compared to the SVM-LSTM, the training and testing RMSE of the LSTM is reduced by 76.62% and 71.18%, respectively. It also decreases the training and testing RMSE of the SVM by 77.06% and 71.81%, respectively. The proposed model can be implemented as reliable solution for accurately predicting solar activities in smart cities. [ABSTRACT FROM AUTHOR]

Connected and autonomous vehicles (CAVs) have witnessed significant attention from industries, and academia for research and developments towards the on-road realisation of the technology. State-of-the-art CAVs utilise existing navigation systems for mobility and travel path planning. However, reliable connectivity to navigation systems is not guaranteed, particularly in urban road traffic environments with high-rise buildings, nearby roads and multi-level flyovers. In this connection, this paper presents TAKEN-Traffic Knowledge-based Navigation for enabling CAVs in urban road traffic environments. A traffic analysis model is proposed for mining the sensor-oriented traffic data to generate a precise navigation path for the vehicle. A knowledge-sharing method is developed for collecting and generating new traffic knowledge from on-road vehicles. CAVs navigation is executed using the information enabled by traffic knowledge and analysis. The experimental performance evaluation results attest to the benefits of TAKEN in the precise navigation of CAVs in urban traffic environments. [ABSTRACT FROM AUTHOR]

A Brain–Computer Interface (BCI), integrated with the Internet of Medical Things (IoMT) and based on electroencephalogram (EEG) technology, allows users to control external devices by decoding brainwave patterns. Advanced deep learning-based BCIs, especially those utilizing sensorimotor rhythms (SMRs), have emerged as direct brain–device communication facilitators. SMRs involve users imagining limb motions to induce specific brain activity changes in the motor cortex. Despite progress, some users struggle with BCIs due to weak signals, individual variability, and limited task applicability. This study introduces an unsupervised EEG preprocessing pipeline for SMR-based BCIs. It evaluates an EEG dataset recorded during finger movements, employing two cleaning methods: an investigator-dependent pipeline and our proposed unsupervised method. Two distinct feature datasets are generated: one from cleaned EEG data processed into spectrogram images using supervised preprocessing, and another from data cleaned using our proposed unsupervised pipeline. The study extensively assesses five transfer learning convolutional neural network (TL-CNN) models for distinguishing Motor Imagery (MI) from finger movements (Mex) using the generated datasets. A novel probability fusion technique is developed to enhance TL-CNN classification in Mex versus MI finger-pinching actions. Comparative results show that the fusion-based method outperforms other state-of-the-art methods when applied to unsupervised EEG data. Specifically, our proposed approach achieves 97.9% accuracy, 93.4% precision, 95% recall, and an F1-score of 93.2%, demonstrating significant progress in distinguishing MI and Mex activities through the use of our unsupervised pre-processing pipeline and fusion-based CNN method. Our findings demonstrate the potential of our approach to serve as a benchmark for the global interdisciplinary research community and enable the development of future more effective and user-friendly real-time BCI systems. • Near real-time EEG preprocessing and probability-based fusion are proposed. • A novel approach to disentangle MI (mental motor acts) from Mex (motor execution). • Hand Movement Classification using a Fusion of CNN architectures in the IoMT. • 64 EEG channels have been used to perform ERS and ERD analyses. • Experiments show good performances on EEG time series and time–frequency datasets. [ABSTRACT FROM AUTHOR]

Federated learning (FL) is a distributed model for deep learning that integrates client-server architecture, edge computing, and real-time intelligence. FL has the capability of revolutionizing machine learning (ML) but lacks in the practicality of implementation due to technological limitations, communication overhead, non-IID (independent and identically distributed) data, and privacy concerns. Training a ML model over heterogeneous non-IID data highly degrades the convergence rate and performance. The existing traditional and clustered FL algorithms exhibit two main limitations, including inefficient client training and static hyperparameter utilization. To overcome these limitations, we propose a novel hybrid algorithm, namely, genetic clustered FL (Genetic CFL), that clusters edge devices based on the training hyperparameters and genetically modifies the parameters clusterwise. Then, we introduce an algorithm that drastically increases the individual cluster accuracy by integrating the density-based clustering and genetic hyperparameter optimization. The results are bench-marked using MNIST handwritten digit dataset and the CIFAR-10 dataset. The proposed genetic CFL shows significant improvements and works well with realistic cases of non-IID and ambiguous data. An accuracy of 99.79% is observed in the MNIST dataset and 76.88% in CIFAR-10 dataset with only 10 training rounds. [ABSTRACT FROM AUTHOR]

Recent technological developments, such as the Internet of Things (IoT), artificial intelligence, edge, and cloud computing, have paved the way in transforming traditional healthcare systems into smart healthcare (SHC) systems. SHC escalates healthcare management with increased efficiency, convenience, and personalization, via use of wearable devices and connectivity, to access information with rapid responses. Wearable devices are equipped with multiple sensors to identify a person's movements. The unlabeled data acquired from these sensors are directly trained in the cloud servers, which require vast memory and high computational costs. To overcome this limitation in SHC, we propose a federated learning-based person movement identification (FL-PMI). The deep reinforcement learning (DRL) framework is leveraged in FL-PMI for auto-labeling the unlabeled data. The data are then trained using federated learning (FL), in which the edge servers allow the parameters alone to pass on the cloud, rather than passing vast amounts of sensor data. Finally, the bidirectional long short-term memory (BiLSTM) in FL-PMI classifies the data for various processes associated with the SHC. The simulation results proved the efficiency of FL-PMI, with 99.67% accuracy scores, minimized memory usage and computational costs, and reduced transmission data by 36.73%. [ABSTRACT FROM AUTHOR]

With the explosive advancements in smartphone technology, video uploading/downloading has become a routine part of digital social networking. Video contents contain valuable information as more incidents are being recorded now than ever before. In this paper, we present a comprehensive survey on information extraction from video contents and forgery detection. In this context, we review various modern techniques such as computer vision and different machine learning (ML) algorithms including deep learning (DL) proposed for video forgery detection. Furthermore, we discuss the persistent general, resource, legal, and technical challenges, as well as challenges in using DL for the problem at hand, such as the theory behind DL, CV, limited datasets, real-time processing, and the challenges with the emergence of ML techniques used with the Internet of Things (IoT)-based heterogeneous devices. Moreover, this survey presents prominent video analysis products used for video forensics investigation and analysis. In summary, this survey provides a detailed and broader investigation about information extraction and forgery detection in video contents under one umbrella, which was not presented yet to the best of our knowledge. [ABSTRACT FROM AUTHOR]

• To develop an efficient network slicing with three phases. • To collect the relevant data attributes of network slicing. • To optimize the weight function GS-DHOA algorithm. Network slicing is predetermined to hold up the diversity of emerging applications with enhanced performance and flexibility requirements in the way of splitting the physical network into numerous logical networks. Consequently, a tremendous data count has been generated with an enormous number of mobile phones due to these applications. This has made remarkable challenges and has a considerable influence on the network slicing performance. This work aims to design an efficient network slicing using a hybrid learning algorithm. Thus, we proposed a model, which involves three main phases: (a) Data collection, (b) Optimal weighted feature extraction (OWFE), and (c) Slicing classification. First, we collected the 5G network slicing dataset, which involves the attributes associated with various network devices like "user device type, duration, packet loss ratio, packet delay budget, bandwidth, delay rate, speed, jitter, and modulation type." Next, we performed the OWFE, in which a weight function is multiplied with the attribute values to have high scale variation. We optimized this weight function by the hybridization of two meta-heuristic algorithms—glowworm swarm optimization and deer hunting optimization algorithm (DHOA)—and named the proposed model glowworm swarm-based DHOA (GS-DHOA). For the given attributes, we classified the exact network slices like "eMBB, mMTC, and URLLC" for each device by a hybrid classifier using deep belief and neural networks. The weight function of both networks is optimized by the GS-DHOA. The experiment results revealed that the proposed model could influence the provision of accurate 5G network slicing. [ABSTRACT FROM AUTHOR]