Sleep is a crucial period of rest necessary for optimal cognitive function, psychological well-being, and execution of everyday tasks. In the field of sleep healthcare, the primary objective is to identify and classify the various sleep states. Implementing sleep state detection in a system is problematic and essential for accurate diagnosis. Our study used an integrated framework to recognize sleep states. The dataset contained approximately eight lakh data points sorted into two groups: onset and wake-up. We successfully deployed a cutting-edge Convolutional Recurrent Transformer (CRT) model for sleep state detection. The training accuracy of our detection model was measured at 97.83%, a constant validation accuracy of 97.07%, and a testing measurement accuracy of 97.23%, were maintained. These scores indicate the model's proficiency in precisely recognizing the sleep states. Our system's detection capabilities demonstrate the ability to identify different sleep states, enhance the accuracy of diagnoses and increase healthcare outcomes in this specialized field.

Code summarization is a process of creating a readable natural language from programming source codes. Code summarization has become a popular research topic for software maintenance, code generation, and code recovery. Existing code summarization methods follow the encoding/decoding approach and use various machine learning techniques to generate natural language from source codes. Although most of these methods are state of the art, it is difficult to understand the complex encoding and decoding process to map the tokens with natural language words. Therefore, these coding and decoding approaches are treated as opaque models (black box). This research proposes explainable AI methods that overcome the black box features for the token mapping in code summarization process. Here, we created an abstract syntax tree (AST) from the tokens of the source code. We then embedded the AST into natural language words using a bilingual statistical probability approach to generate possible statistical parse trees. We applied a page rank algorithm among the parse trees to rank the trees. From the best-ranked tree, we generate the comment for the corresponding code snippet. To explain our code generation method, we used Takagi–Sugeno fuzzy approach, layerwise relevance propagation and a hidden Markov model. These approaches make our method trustworthy and understandable to humans to understand the process of source code token mapping with natural language words.

Abstract Today, the development of urbanization and increasing the number of vehicles has resulted in displeased consequences like traffic congestion and vehicle queuing. The vast majority of countries in the world encounter the challenge of the explosive rise in traffic demand. In this regard, it is necessary to meet traffic demand in transport networks, especially in metropolitans. In traffic management and shortening the trip duration, traffic lights on the signalized intersections play an essential role in urban pathways. This work provides a multi‐criteria decision‐making method for optimum traffic light control in an isolated corner. The main idea involves establishing a set of sub‐optimal solutions for traffic light timing and selecting the best one among the diverse solutions. We have mathematically modelled the problem as an optimization problem to achieve an optimal solution with less waiting time for vehicles in intersections and the lowest cost. Genetic algorithm (GA) and Teaching‐Learning‐based Optimization (TLBO) are utilized for each phase to create a set of suitable timing scenarios. The Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method is used to identify the best scenario, considering both waiting vehicles and traffic capacity as decision criteria. Its efficiency has been demonstrated over three different traffic volumes. Also, in a real‐world implementation, its practical capability has been approved at a crossroads in Mashhad, Iran. The simulations indicate the improvement in the number of vehicles waiting behind the crossroad and the traffic capacity by 10% and 6.76% compared to the existing signal timing of the studied intersection, respectively.

The current era witnesses the notable transition of society from an information-centric to a human-centric one aiming at striking a balance between economic advancements and upholding the societal and fundamental needs of humanity. It is undeniable that the Internet of Things (IoT) and artificial intelligence (AI) are the key players in realizing a human-centric society. However, for society and individuals to benefit from advanced technology, it is important to gain the trust of human users by guaranteeing the inclusion of ethical aspects such as safety, privacy, nondiscrimination, and legality of the system. Incorporating explainable AI (XAI) into the system to establish explainability and transparency supports the development of trust among stakeholders, including the developers of the system. This article presents the general class of vulnerabilities that affect IoT systems and directs the readers’ attention toward intrusion detection systems (IDSs). The existing state-of-the-art IDS system is discussed. An attack model modeling the possible attacks is presented. Furthermore, since our focus is on providing explanations for the IDS predictions, we first present a consolidated study of the commonly used explanation methods along with their advantages and disadvantages. We then present a high-level human-inclusive XAI framework for the IoT that presents the participating components and roles. We also hint upon a few approaches to upholding safety and privacy using XAI that we will be taking up in our future work. An attack model based on the study of possible attacks on the system is also presented in the article. The article also presents guidelines to choose a suitable XAI method and a taxonomy of explanation evaluation mechanisms, which is an important yet less visited aspect of explainable AI.

Abstract Unmanned aerial vehicle based precision agriculture is a predominant research area. The modern flying ad‐hoc network leverages the advanced low latency vehicular communication and intelligent computing paradigms that help the ecosystem to grow up to the next level. In this work, we propose an ecosystem for precision agriculture that leverages the use of the opportunistic MQTT protocol in an edge‐enabled intelligent drone network for sensing and performing crop prediction using an intelligent ensemble machine learning model. The proposed approach leverages the edge computing system that requires low energy devices and also exploits the ultra‐low latency opportunistic message transfer methodology. The experimental results show the maximum of 0.9 message delivery ratio and a minimum of 600 ms latency is achieved by opportunistic MQTT protocol in an ultra‐low latency sparse network scenario. A weighted ensemble model is deployed onto the edge enabled devices or the drones. An accuracy of 96.5% is achieved in predicting the type of crops that can be grown in the soil about the selected area of interest.

Model inversion attack (MIA) is a cyber threat with an increasing alert even for deep-learning-based recognition systems (DLRSs). By targeting a DLRS under a scenario of attacker access to the model structure and parameters, MIA generates a data clone for a certain targeted class label. To avoid the possible threats of such MIA-generated data clones, this research work proposes a social IoT approach to a collaborative cyber-defense among the online recognition systems (RSs) sharing the targeted class label. Since, the generation of an MIA-clone is by targeting an RS model and using its structure, parameters, and class labels output scores in an iterative optimization process, the generated clone is partially inherent to the targeted model. Thus, it is expected for an MIA-clone to show a different performance on a secondary RS wherein the same targeted class label is included. It is because, in the MIA generation of the clone, not only the targeted class label but also other class labels, and model parameters and structure affect the process, while the second model has just the targeted class label in common with the target model. Deploying the Social Internet of Recognition Systems (SIoRS), the proposed technique utilizes a collaborative recognition by SIoRC which plays the role of a complementary recognition besides the targeted RS. The recognition output by the targeted RS is further verified by the SIoRS complementary recognition result. To avoid the MIA-targeted data clones, the verification of recognition is by the log-likelihood ratio test between the targeted RS and the SIoRS complementary recognition confidence scores. The proposed technique is evaluated by statistical analysis on deep face RSs in 10000 Monte Carlo runs for each of the conventional, dc-generative adversarial network (GAN) and$\alpha $-GAN integrated MIA techniques in targeting two different user identities. The$Z$scores of the fitted normal distribution of the log-likelihood ratios indicate almost 100% detection rate of clones generated by conventional MIA and 95.23% and 86% of clones, respectively, generated by DC-GAN and$\alpha $-GAN integrated deep MIA techniques.

Abstract Predicting patients with major depression (MDD) is currently a difficult task. Magnetic resonance imaging (MRI) data analysis may provide insight into individual patient responses, allowing for more customized treatment decisions. Due to the absence of brain MRI data for MDD patients, a transfer learning (TL) method developed is used using calculation criteria. Combining an Inception‐v3 neural network with a typical pre‐trained neural network, the move learning‐based Inception‐v3 was proposed for the classification of MDD MRI datasets. An experiment was performed on the classification of eight semantic emotions (defined by IMAPS). Compared to other methods, the proposed method performs high efficiency for 90–10% and 80–20% (positive and negative classes), normal (N), unnormal (UN), and average/total sets, and for 70–30%, accuracy (A) is 92.90%, area under the curve (AUC) is 94.23%, and average precision score (APS) is 95.75%. Individual patients' responses to emotional stimulation can be predicted using the proposed methods, which can provide guidance in diagnosis and prognosis.

Smart monitoring of off-road vehicles is cursed by their complex and expensive IoT sensors technologies. High dependence on the cloud/fog computation, availability of the network, and expert knowledge make it handicap in the rural off-network areas. Use of edge devices, such as smartphones, attributed by computation capabilities is the solution that is yet to be developed at commercial level (Fawwaz and Chung 2020) and (Zhengweiet al., 2021). Additionally, the user's growing demand for economic and user-friendly technology motivates to shift from costly and complex sensors to economic. In this article, we present the hybridized computational intelligence methodology to develop an edge-device-enabled AI technology for health monitoring and diagnosis (HM&D) of the off-road vehicles, taking use of super economic microphones as sensors. Smartphones are benefited by integrated microphones, and thus, the App-based developed technology is generalized for all vehicles from old to new. Enhanced selection and log-scaled mutation genetic algorithms is used to evolve the structure of the artificial neural network toward an optimally lightweight structure. Each evolved lightweight ANN structure is trained by scaled conjugate gradient back-propagation training algorithm to optimize corresponding weights and biases. The comparative results with currently reported genetic algorithms for edge computation prove it a breakthrough technology for edge-device-enabled HM&D of off-road vehicles (Yanet al., 2020).

Telecommunications systems with Multi-Input Multi-Output (MIMO) structure using Orthogonal Frequency Division Modulation (OFDM) have great potential for efficient application to a network of Internet of Things (IoT) at a high data rate. When the IoT network is among the underwater sensory devices known as the Internet of Underwater Things (IoUT), the electromagnetic wave cannot play the role of baseband signal due to rapid fall-off inside the water. Thus, acoustic OFDM is a reliable replacement for conventional OFDM inside the water. A blind structure for MIMO acoustic OFDM using Independent Component Analysis (ICA) brings even further advantages in data rate and energy consumption by avoiding the required pilot and preamble data. This research work presents a blind MIMO Acoustic OFDM blind transceiver for IoUT based on Probabilistic Stone’s Blind Source Separation (PS-BSS). The proposed technique has multiple times lower complexity compared to the ICA-based technique while maintaining a comparable efficiency. As observed in the results carried out with 100 Monte Carlo runs of transmission of random data bits over a highly sparse channel that is the common case of an underwater environment, the proposed PS-BSS-based technique dominates the ICA-based one, and as the sparseness of the channel decreases, its efficiency is comparable to the ICA-based technique. Thus, in the case of a highly sparse channel, the proposed technique is superior in both aspects of efficiency and complexity, while over lower sparseness, due to its comparative efficiency, it can be hired as an optimum technique fulfilling a fair tradeoff between efficiency and complexity.

Analysis of irregularities in Covid-19 data could open a new window to learn more about the unprecedented problems of the current global pandemic. Of many, radiographs and clinical records are reliable sources for viral infection investigation and treatment planning. Clinical records help track the Covid-19 pandemic. In this paper, we present a Spike Neural Network (SNN) with supervised synaptic learning to detect abnormalities in Chest X-rays (CXRs) In other words, the proposed SNN can distinguish Covid-19 positive cases from healthy ones. In our decision-making procedure, we introduce clinical practice so Explainable AI (XAI) is possible to carry out. In addition, Support Vector Machine (SVM) with local interpretable model-agnostic explanation (LIME) provides reliable analysis of abnormalities in Covid-19 clinical data.

Wearable on body sensors have been employed in many applications including ambulatory monitoring and pervasive computing systems. In this work, a wearable assistant has been created for people suffering from Parkinson’s disease (PD), specifically with the freezing of gait (FoG) symptom. Wearable accelerometers were placed on the person’s body and used for movement measure. When FoG is detected, a rhythmic audio signal was given from the wearable assistant to motivate the wearer to continue walking. Long-term monitoring results in collecting huge amounts of complex raw data; therefore, data analysis becomes impractical or infeasible resulting in the need for data reduction. In the present study, discrete wavelet transform (DWT) has been used to extract the main features inherent in the key movement indicators for FoG detection. The discrimination capacities of these features were assessed using (i) support vector machine using a linear kernel function and (ii) artificial neural network with a two-layer feed-forward with hidden layer of 20 neurons that trained with conjugate gradient back-propagation. Using these two different machine learning techniques, we were capable of detecting FoG with an accuracy of 87.50% and 93.8%, respectively. Additionally, the comparison between the extracted features from DWT coefficients with those using fast Fourier transform established accuracies of 93.8% and 81.3%, respectively. Finally, the discriminative features extracted from DWT yield to a robust multidimensional classification model compared to models in the literature based on a single feature. The work presented paves the way for reliable, real-time wearable sensors to aid people with PD.

Social media conveys a reachable platform for users to share information. The inescapable practice of social media has produced remarkable volumes of social data. Social media gathers the data in both structured-unstructured and formal-informal ways as users are not concerned with the exact grammatical structure and spelling when interacting with each other by means of various social networking websites (Twitter, Facebook, YouTube, LinkedIn, etc.). People are increasingly involved in and dependent on social media networks for data, news and opinions of other handlers on a variety of topics. The strong dependence on social media network sites contributes to enormous data generation characterized by three issues: scale, noise, and variety. Such problems also hinder social network data to be evaluated manually, resulting in the correct use of statistical analytical methods. Mining social media data can extract significant patterns that can be advantageous for consumers, users, and business. Pattern mining offers a wide variety of methods to detect valuable knowledge from huge datasets, such as patterns, trends, and rules. In this work, data was collected comprised of users’ opinions and sentiments and then processed using a significant number of pattern mining methods. The results were then further analyzed to attain meaningful information. The aim of this paper is to deliver a summary and a set of strategies for utilizing the ubiquitous pattern mining approaches, and to recognize the challenges and future research guidelines of dealing out social media data.

Glaucoma is a major public health problem that can lead to an optic nerve lesion, requiring systematic screening in the population over 45 years of age. The diagnosis and classification of this disease have had a marked and excellent development in recent years, particularly in the machine learning domain. Multimodal data have been shown to be a significant aid to the machine learning domain, especially by its contribution to improving data driven decision-making.Solving classification problems by combinations of classifiers has made it possible to increase the robustness as well as the classification reliability by using the complementarity that may exist between the classifiers. Complementarity is considered a key property of multimodality. A Convolutional Neural Network (CNN) works very well in pattern recognition and has been shown to exhibit superior performance, especially for image classification which can learn by themselves useful features from raw data. This article proposes a multimodal classification approach based on deep Convolutional Neural Network and Support Vector Machine (SVM) classifiers using multimodal data and multimodal feature for glaucoma diagnosis from retinal fundus images from RIM-ONE dataset. We make use of handcrafted feature descriptors such as the Gray Level Co-Occurrence Matrix, Central Moments and Hu Moments to co-operate with features automatically generated by the CNN in order to properly detect the optic nerve and consequently obtain a better classification rate, allowing a more reliable diagnosis of glaucoma.The experimental results confirm that the combination of classifiers using a new hybrid fusion approach and the BWWV technique is better than learning classifiers separately. The proposed method provides a computerized diagnosis system for glaucoma disease with impressive results comparing them to the main related studies that allow us to continue in this research path.

Domestic and industrial pollutions affected the water quality to a greater extent. Polluted water became a major reason behind several community diseases, mainly in undeveloped and developing countries. The public health condition is deteriorating and putting an extra burden of countermeasures to prevent such water borne diseases from spreading. Detecting the drinking water quality can prevent such scenarios prior to the critical stage. Recent research works have achieved reasonable success in predicting the water quality. However, the accuracy levels of already proposed models are to be improved, keeping in mind the sensitivity of the problem domain. In the current work, multi-objective genetic algorithm was employed to train the artificial neural network (NN-MOGA) to improve its performance over its traditional counterparts. The proposed model gradually minimizes two different objective functions; namely the root mean square error (RMSE) and Maximum Error in order to find the optimal weight vector for the artificial neural network (ANN). The proposed model was compared with three other, well established models namely NN-GA (ANN trained with Genetic Algorithm), NN-PSO (ANN trained with Particle Swarm Optimization) and SVM in terms of accuracy, precision, recall, F-Measure, Matthews correlation coefficient (MCC) and Fowlkes-Mallows index (FM index). The simulation results established superior accuracy of NN-MOGA over the other models.

In the current digitalized era, large datasets play a vital role in features extractions, information processing, knowledge mining and management. Sometimes, existing mining approaches are not sufficient to handle large volume of datasets. Biological data processing also suffers for the same issue. In the present work, a classification process is carried out on large volume of exons and introns from a set of raw data. The proposed work is designed into two parts as pre-processing and mapping-based classification. For pre-processing, three filtering techniques have been used. However, these traditional filtering techniques face difficulties for large datasets due to the long required time during large data processing as well as the large required memory size. In this regard, a mapping-based neural skyline filtering approach is designed. Randomized algorithm performed the mapping for large volume of datasets based on objective function. The objective function determines the randomized size of the datasets according to the homogeneity. Around 200 million DNA base pairs have been used for experimental analysis. Experimental result shows that mapping centric filtering outperforms other filtering techniques during large data processing.

The three-dimensional (3D) mesh is moderately novel media type that realizes a rising success in various applications through data transfer via the Internet, which requires security approaches. Technological copyright protection of digital contents has become a challenging task in the current digital epoch. In this work, a robust watermarking algorithm of polygonal meshes for copyright protection purposes is proposed. The watermark insertion was achieved by quantization of the vertex norms variance in order to insert the watermark bits. In addition, this method is based on a blind detection scheme, so the watermark can be extracted without referring to the original mesh. The experimental results established the quality of the watermarked object as well as the inserted watermark robustness against various types of attacks, which were evaluated to prove the validity of the proposed algorithm. The results proved the proposed method efficiency in terms of robustness and imperceptibility against several signal processing distortions. A comparison with other reported method with similar purposes is also provided. The comparison depicted the outstanding robustness of the proposed method compared to the other reported method.

Face recognition is an importance step which can affect the performance of the system. In this paper, the authors propose a novel Max-Min Ant System algorithm to optimal feature selection based on Discrete Wavelet Transform feature for Video-based face recognition. The length of the culled feature vector is adopted as heuristic information for ant's pheromone in their algorithm. They selected the optimal feature subset in terms of shortest feature length and the best performance of classifier used k-nearest neighbor classifier. The experiments were analyzed on face recognition show that the authors' algorithm can be easily implemented and without any priori information of features. The evaluated performance of their algorithm is better than previous approaches for feature selection.

The proliferation of information and communication technology has made exchange of information easier than ever. Security, Duplication and manipulation of information in such a scenario has become a major challenge to the research community round the globe. Digital watermarking has been found to be a potent tool to deal with such issues. A secure and robust image watermarking scheme based on DC coefficient modification in pixel domain and chaotic encryption has been presented in this paper. The cover image has been divided into 8×8 sub-blocks and instead of computing DC coefficient using Discrete Cosine Transform (DCTI, the authors compute DC coefficient of each block in spatial domain. Watermark bits are embedded by modifying DC coefficients of various blocks in spatial domain. The quantum of change to be brought in various pixels of a block for embedding watermark bit depends upon DC coefficient of respective blocks, nature of watermark bit (0 or 1) to be embedded and the adjustment factor. The security of embedded watermark has been taken care of by using chaotic encryption. Experimental investigations show that besides being highly secure the proposed technique is robust to both signal processing and geometric attacks. Further, the proposed scheme is computationally efficient as DC coefficient which holds the watermark information has been computed in pixel domain instead of using DCT on an image block.

Summary Software development life cycle is a structured process, including the definition of user requirements specification, the system design, and programming. The design task comprises the transfer of natural language specifications into models. The class diagram of Unified Modeling Language has been considered as one of the most useful diagrams. It is a formal description of user's requirements and serves as inputs to the developers. The automated extraction of UML class diagram from natural language requirements is a highly challenging task. This paper explains our vision of an automated tool for class diagram generation from user requirements expressed in natural language. Our new approach amalgamates the statistical and pattern recognition properties of natural language processing techniques. More than 1000 patterns are defined for the extraction of the class diagram concepts. Once these concepts are captured, an XML Metadata Interchange file is generated and imported with a Computer‐Aided Software Engineering tool to build the corresponding UML class diagram. Copyright © 2015 John Wiley & Sons, Ltd.

The most computationally expensive operation in entire video compression process is Motion Estimation. The challenge is to reduce the computational complexity and time of Exhaustive Search Algorithm without losing too much quality at the output. The proposed work is to implement a novel block matching algorithm for Motion Vector Estimation which performs better than other conventional Block Matching Algorithms such as Three Step Search (TSS), New Three Step Search (NTSS), and Four Step Search (FSS) etc.

At present most of the hospitals and diagnostic centers globally, use wireless media to exchange biomedical information for mutual availability of therapeutic case studies. The required level of security and authenticity for transmitting biomedical information through the internet is quite high. Level of security can be increased; authenticity of the information can be verified and control over the copy process can be ascertained by adding watermark as “ownership” information in multimedia content. In this proposed method different types of gray scale biomedical images can be used as added ownership (watermark) data. Electrooculography is a medical test used by the ophthalmologists for monitoring eyeball movement in Rapid Eye Movement (REM) and non-REM sleep, to detect the disorders of human eyes and to measure the resting potential of the eye. In this present work 1-D EOG signal is transformed into 2-D signal. DWT, DCT, SVD are applied on the transformed 2D signal to embed watermark in it. Extraction of watermark image is done by applying inverse DWT, inverse DCT and SVD. The Peak Signal to Noise Ratio (PSNR) of the original EOG signal vs. watermarked signal and the correlation value between the original and extracted watermark image are calculated to prove the efficacy of the proposed method.

The present medical era has seen quite a considerable amount of work been done in tele-monitoring that involves transmission of biomedical signals through wireless media. Exchange of information amongst various hospital systems and medical centers require high level of reliability and security. Signal integrity can be verified, authenticity and achieved control over the copy process can be proved by adding watermark in the original information as multimedia content. Electroencephalography (EEG) is a medical test that records the electrical activity originating from the brain. In this present work, Alattar's Method is used for watermark insertion and extraction in an EEG signal without devalorizing its diagnostic parameters. In the second part of the paper, different features in the time domain and the spatial domain are obtained from the original EEG signal, watermarked EEG signal, and the recovered EEG signal.