Over the last few decades, there has been an increase in the probability of occurrence of wildfires. Also known as bushfires, the catastrophe can be attributed to climate changes and extreme weather conditions. Australia's dry and warm climate makes it prone to wildfires, which puts the ecosystem at risk and decreases the forest area. Hence, it is necessary to reduce bushfire risk by monitoring their intensity. The availability of remotely sensed data enables us to analyze wildfires, explore and discover the patterns, and help provide real-time warnings. This work examined the forest fire data during 2018–2020, considering parameters like Brightness (Prediction) and Fire Radiative Power (Classification). The analysis is conducted using several machine learning algorithms like Random Forest (RF), Decision Trees (DT), k-Nearest Neighbors (kNN), eXtreme Gradient Boosting (xGB), Artificial Neural Networks (ANN), Convolution Neural Networks (CNN), etc. The prediction models were evaluated using Mean Squared Error (MSE), Root Mean Square Error (RMSE), R squared (R2), and Mean Absolute Error (MAE). In contrast, the classification models were assessed based on accuracy, precision, recall, and F-1 score. Our study shows that the RF model is the best prediction model, and the ANN model is the best classification model compared to the baseline models.

The Internet of Things (IoT) is being prominently used in smart cities and a wide range of applications in society. The benefits of IoT are evident, but cyber terrorism and security concerns inhibit many organizations and users from deploying it. Cyber-physical systems that are IoT-enabled might be difficult to secure since security solutions designed for general information/operational technology systems may not work as well in an environment. Thus, deep learning (DL) can assist as a powerful tool for building IoT-enabled cyber-physical systems with automatic anomaly detection. In this paper, two distinct DL models have been employed i.e., Deep Belief Network (DBN) and Convolutional Neural Network (CNN), considered hybrid classifiers, to create a framework for detecting attacks in IoT-enabled cyber-physical systems. However, DL models need to be trained in such a way that will increase their classification accuracy. Therefore, this paper also aims to present a new hybrid optimization algorithm called “Seagull Adapted Elephant Herding Optimization” (SAEHO) to tune the weights of the hybrid classifier. The “Hybrid Classifier + SAEHO” framework takes the feature extracted dataset as an input and classifies the network as either attack or benign. Using sensitivity, precision, accuracy, and specificity, two datasets were compared. In every performance metric, the proposed framework outperforms conventional methods.

Chronic Kidney Disease (CKD) pose a global health challenge due to their increasing incidence and delayed detection, compounded by the burden they place on healthcare systems worldwide. This surge is particularly pronounced in both industrialized and emerging nations, fueled by the rising prevalence of conditions such as diabetes, hypertension, and obesity. While Artificial Intelligence (AI) shows promise for early CKD detection, issues with data security, model transparency, and decision-making trust have created uncertainty among both patients and healthcare providers. To tackle these issues, a decentralized collaborative learning framework utilizing explainable AI to detect CKD is proposed. This approach combines the privacy-preserving nature of blockchain technology with the interpretability of explainable AI models.Areal-time environmentwas set up to conduct these experiments, simulating a network of healthcare providers collaborating on CKD detection while maintaining patient privacy.

The rapid rise of technological advancements led to the increased consumption of electronic gadgets. This change expedited the requirement for sustainable technologies to meet the growing consumer requirements with minimum computational costs. Nowadays, video content shares a large proportion of the internet bandwidth. Object Detection from the videos is essential in various real-time applications. Traditionally, the videos are decoded to the raw format for detection tasks. This analytics process can be more efficient if the detection tasks are carried out from compressed video formats instead of raw video. The compressed format of the videos, produced by modern deep learning-based approaches, contains both semantic and motion information in easily consumable formats. Based on the same notion, a video compression cum object detection network has been proposed in this paper, which consumes the compressed videos for carrying out detection tasks. The proposed network comprises an already-designed video compression network, which has been extended to incorporate object detection capabilities. The proposed network has been experimented with a standard ImageNet VID dataset, and the results show fast and efficient object detection from the compressed videos. Coupled with temporal features, the proposed model achieves significantly better mAP of 44.3 w.r.t. 36.7 and 39.6 from YOLOv5-s and YOLOX-s models, respectively. The comparative results have shown incremental improvement in the detection tasks from the compressed videos, making it sustainable for its application in modern lightweight consumer electronic devices.

The swift proliferation of the Internet of Things (IoT) devices in smart city infrastructures has created an urgent demand for robust cybersecurity measures. These devices are susceptible to various cyberattacks that can jeopardize the security and functionality of urban systems. This research presents an innovative approach to identifying anomalies caused by IoT cyberattacks in smart cities. The proposed method harnesses federated and split learning and addresses the dual challenge of enhancing IoT network security while preserving data privacy. This study conducts extensive experiments using authentic datasets from smart cities. To compare the performance of classical machine learning algorithms and deep learning models for detecting anomalies, model effectiveness is assessed using precision, recall, F-1 score, accuracy, and training/deployment time. The findings demonstrate that federated learning and split learning have the potential to balance data privacy concerns with competitive performance, providing robust solutions for detecting IoT cyberattacks. This study contributes to the ongoing discussion about securing IoT deployments in urban settings. It lays the groundwork for scalable and privacy-conscious cybersecurity strategies. The results underscore the vital role of these techniques in fortifying smart cities and promoting the development of adaptable and resilient cybersecurity measures in the IoT era.

In numerous scientific disciplines and practical applications, addressing optimization challenges is a common imperative. Nature-inspired optimization algorithms represent a highly valuable and pragmatic approach to tackling these complexities. This paper introduces Dendritic Growth Optimization (DGO), a novel algorithm inspired by natural branching patterns. DGO offers a novel solution for intricate optimization problems and demonstrates its efficiency in exploring diverse solution spaces. The algorithm has been extensively tested with a suite of machine learning algorithms, deep learning algorithms, and metaheuristic algorithms, and the results, both before and after optimization, unequivocally support the proposed algorithm’s feasibility, effectiveness, and generalizability. Through empirical validation using established datasets like diabetes and breast cancer, the algorithm consistently enhances model performance across various domains. Beyond its working and experimental analysis, DGO’s wide-ranging applications in machine learning, logistics, and engineering for solving real-world problems have been highlighted. The study also considers the challenges and practical implications of implementing DGO in multiple scenarios. As optimization remains crucial in research and industry, DGO emerges as a promising avenue for innovation and problem solving.

Optimizing algorithmic efficiency in today’s computing landscape remains a pressing concern across various fields. This study conducts an in-depth analysis to address this challenge by leveraging quantum-inspired optimization techniques (QIOT). Specifically, the research focuses on the integration of QIOT with deep neural networks (DNNs) to enhance model performance across diverse datasets. Through extensive experimentation, ten swarm-intelligence-based optimization techniques are investigated, including Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Grey Wolf Optimizer (GWO), Firefly Algorithm (FO), Bee Colony Optimization (BCO), Cuckoo Search (CS), Moth Flame Optimization (MFO), Flower Pollination Algorithm (FPA), Whale Optimization Algorithm (WOA), and Bat Algorithm (BA). These techniques were chosen for their decentralized, collective behavior, making them particularly suitable for integration with quantum-inspired methods. Evaluation metrics such as accuracy, precision, recall, F1 score, and training time are employed in this study. Furthermore, comparative analyses demonstrate the superiority of QIOT over traditional optimization methods in enhancing algorithmic efficiency. By deploying three different datasets, the study observes a reduction in training time by 27

The sudden unexpected rise in monkeypox cases worldwide has become an increasing concern. The zoonotic disease characterized by smallpox-like symptoms has already spread to nearly twenty countries and several continents and is labeled a potential pandemic by experts. monkeypox infections do not have specific treatments. However, since smallpox viruses are similar to monkeypox viruses administering antiviral drugs and vaccines against smallpox could be used to prevent and treat monkeypox. Since the disease is becoming a global concern, it is necessary to analyze its impact and population health. Analyzing key outcomes, such as the number of people infected, deaths, medical visits, hospitalizations, etc., could play a significant role in preventing the spread. In this study, we analyze the spread of the monkeypox virus across different countries using machine learning techniques such as linear regression (LR), decision trees (DT), random forests (RF), elastic net regression (EN), artificial neural networks (ANN), and convolutional neural networks (CNN). Our study shows that CNNs perform the best, and the performance of these models is evaluated using statistical parameters such as mean absolute error (MAE), mean squared error (MSE), mean absolute percentage error (MAPE), and R-squared error (R2). The study also presents a time-series-based analysis using autoregressive integrated moving averages (ARIMA) and seasonal auto-regressive integrated moving averages (SARIMA) models for measuring the events over time. Comprehending the spread can lead to understanding the risk, which may be used to prevent further spread and may enable timely and effective treatment.

Over the last few decades, the expansion of technology and the internet has led to the number of users proliferating on social media, with a simultaneous increase in hate speech. A critical concern is, hate speech is not only responsible for igniting violence and spreading hatred, but its detection also requires a considerable amount of computing resources and content monitoring by human experts and algorithms. While the research is an active area, and several artificial intelligence techniques have been proposed in the past to address the concern, the rise in the number of petabytes of the content generated calls for methods that will exhibit improved performance and reduced model development time. We propose a transfer learning approach for detecting hate and offensive speech on social media that deploys a pre-trained model for data analysis thereby promoting model reusability. We propose two transfer learning models, i.e. Google’s Word2vec model using LSTM and GloVe Model using LSTM for the same and compare the performance of our proposed model against unigram and bigram language models for Naive Bayes (NB), Decision Trees (DT), and Support Vector Machines (SVM), which are also the baseline algorithms considered for analysis. The performance of the proposed models for classifying hate speech, offensive speech, and neutral speech is validated using metrics such as precision, recall, F-1 score, and support. The overall performance of the models across multiple datasets has been evaluated with respect to accuracy. In-depth experimental analysis and results depict that the proposed model is significantly robust for detecting hateful and offensive speech and also performs better than the considered baseline algorithms.

In recent years, the proliferation of fake accounts on social media has become a significant concern for individuals, organizations, and society. Fake accounts play an important role in spreading fake news, rumors, spam, unethical harassment, and other mischievous motives on social media platforms. Detecting such accounts manually is time-consuming and challenging, especially with the increasing sophistication of the methods used to create them. Therefore, there is a need for automated approaches to detect these fake accounts. This article aims to reveal fake accounts and their role on social media platforms. This article has explored various methodologies to detect fake accounts from social platforms, which will help prevent cybercrime. We have also proposed a generalized deep learning (DL) model to detect such accounts on social media using multimodal data. The proposed architecture uses a combination of textual, visual, and network-based features to capture the various characteristics of fake accounts. Specifically, we use a deep neural network that combines convolutional neural networks (CNNs) for visual data, long short-term memory (LSTM) networks for textual data, and convolutional graph networks (GCNs) for network-based data. We evaluated our model on a publicly available dataset of Twitter accounts and achieved state-of-the-art performance in detecting fake accounts, with an $F$ 1 score of 0.96. We also conducted experiments to show the effectiveness of each feature and the combination of the three features.

Autism spectrum disorder (ASD) has been associated with conditions like depression, anxiety, epilepsy, etc., due to its impact on an individual’s educational, social, and employment. Since diagnosis is challenging and there is no cure, the goal is to maximize an individual’s ability by reducing the symptoms, and early diagnosis plays a role in improving behavior and language development. In this paper, an autism screening analysis for toddlers and adults has been performed using fair AI (feature engineering, SMOTE, optimizations, etc.) and deep learning methods. The analysis considers traditional deep learning methods like Multilayer Perceptron (MLP), Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM), and also proposes two hybrid deep learning models, i.e., CNN–LSTM with Particle Swarm Optimization (PSO), and a CNN model combined with Gated Recurrent Units (GRU–CNN). The models have been validated using multiple performance metrics, and the analysis confirms that the proposed models perform better than the traditional models.

Over the last few decades, there has been an increase in the probability of occurrence of wildfires. Also known as bushfires, the catastrophe can be attributed to climate changes and extreme weather conditions. Australia’s dry and warm climate makes it prone to wildfires, which risks the ecosystem and decreases the forest area. Hence it is necessary to reduce bushfire risk by monitoring their intensity. The availability of remotely sensed data enables us to analyse wildfires, explore and discover patterns, and help provide real-time warnings. This paper examines the forest fire data from 2018-2020, considering parameters like Brightness (Prediction) and Fire Radiative Power (Classification). The analysis is conducted using several machine learning algorithms like Random Forest (RF), Decision Trees (DT), K-Nearest Neighbors (kNN), eXtreme Gradient Boosting (xGB), Artificial Neural Networks (ANN), Convolution Neural Networks (CNN), etc. The prediction models are evaluated using Mean squared error (MSE), Root Mean Square Error (RMSE), R squared (R2), and Mean Absolute Error (MAE). In contrast, the classification models are evaluated using accuracy, precision, recall, and F-1 score. Our study shows that the RF model is the best prediction model, and the ANN model is the best classification model compared to the baseline models.

Respiration is a necessary process for producing energy and maintaining normal bodily functioning in all living organisms. The respiratory system and breathing frequency change per the body’s needs in response to different physical activities, such as running, and to emotional states such as joy and fear. Therefore, this work presents a simulation-based Internet-of-Things (IoT) sensor module using thermistors to estimate the respiration rate (RR) of a human subject and to compare the temperature at the time of breathing. The circuit diagram of the explained sensor was designed and validated using simulations with Proteus software. The results are presented in the form of graphs, comparing resistance and voltage. Specifically, the resistance varies with the temperature near the thermistor, subsequently changing the voltage, which is converted into a digital value to calculate the RR and length of respiration. The main focus of the proposed work for developing this basic circuit is to observe the breathing pattern of a person. From the breathing pattern, many physical activities can be predicted, like he is in a consciousness stage or unconsciousness stage, because every physical activity makes an impact on the breathing pattern. The obtained information can be stored and communicated across the cloud, from which the automated respiratory tracking system can manage and monitor accidental situations. In the case of an emergency, the system sends an alert so that necessary steps can be taken to help the user. Finally, we discuss some applications of the proposed module, specifically for reducing accidental deaths.

Due to data collection, there is a potential risk concerning security and privacy, so IoT reliability and survivability are of utmost concern. In this paper, we address the concern using two methods. The first method is device identification, which uses an extensive set of machine learning algo-rithms for identifying IoT devices. The algorithms include Logistic Regression, K-Nearest Neighbour, Support Vector Classifier, Random Forest, Gradient Boosting, AdaBoost, Light Gradient Boosting Machine, Extreme Gradient Boosting Convolution Neural Networks, and Long Short Term Memory are used for device identification. The performance of these models has been evaluated using multiple statistical measures, and weobserve that LSTM outperforms all other baseline models. The second method proposed for ensuring the survivability of theIoT environ-ment is a blockchain-based architecture for smart homes to ensuretransparency and data protection.

The Internet of Things (IoT) is connecting more devices every day. Security is critical to ensure that the devices operate in a trusted environment. The lack of proper IoT security encourages cybercriminals to target many smart devices across the network and gain sensitive information. Distributed Denial of Service (DDoS) attacks are common in the IoT infrastructure and involve hijacking IoT devices to consume resources and interrupt services. This may specifically vandalize the application running the service that the end users are trying to access (application layer DDoS attacks) or flood the network bandwidth leading to network failure (software defined network DDoS attacks). This article proposes a hybrid attention‐based bidirectional long short term memory (LSTM) with convolutional neural networks (CNN) to identify DDoS attacks in the application layer and SDN. We deploy several other machine learning models like logistic regression, decision trees, random forests, support vector machines, K‐nearest neighbors, extreme gradient boosting, artificial neural networks, CNN, LSTM, CNN‐LSTM to evaluate the performance of our proposed model. The evaluation metrics considered for the study are accuracy, precision, recall, and F‐1 score. The experimental analysis on multiple datasets exhibits that the proposed model performs the classification efficiently with an accuracy of 99.74% and 99.98%.

IoT devices collect time-series traffic data, which is stochastic and complex in nature. Traffic flow prediction is a thorny task using this kind of data. A smart traffic congestion prediction system is a need of sustainable and economical smart cities. An intelligent traffic congestion prediction model using Seasonal Auto-Regressive Integrated Moving Average (SARIMA) and Bidirectional Long Short-Term Memory (Bi-LSTM) is presented in this study. The novelty of this model is that the proposed model is hybridized using a Back Propagation Neural Network (BPNN). Instead of traditionally presuming the relationship of forecasted results of the SARIMA and Bi-LSTM model as a linear relationship, this model uses BPNN to discover the unknown function to establish a relation between the forecasted values. This model uses SARIMA to handle linear components and Bi-LSTM to handle non-linear components of the Big IoT time-series dataset. The "CityPulse EU FP7 project" is a freely available dataset used in this study. This hybrid univariate model is compared with the single ARIMA, single LSTM, and existing traffic prediction models using MAE, MSE, RMSE, and MAPE as evaluation indicators. This model provides the lowest values of MAE, MSE, RMSE, and MAPE as 0.499, 0.337, 0.58, and 0.03, respectively. The proposed model can help to predict the vehicle count on the road, which in turn, can enhance the quality of life for citizens living in smart cities.

The sudden outbreak of the novel coronavirus (nCoV-19, COVID-19) and its rampant spread led to a significant number of people being infected worldwide and disrupted several businesses. With most of the countries imposing serious lockdowns due to the increasing number of fatalities, the social lives of millions of people were affected. Although the lockdown led to an increase in network activities, online shopping, and social network usage, it also raised questions On the mental wellness of society. Interestingly, excessive usage of social networks also witnessed humor traveling across the Internet in the form of Internet Memes during the lockdown period. Humor is known to affect our well-being, decision-making, and psychological systems. In this paper, we have analyzed the Internet Meme activity in Social Networks during the COVID-19 Lockdown period. As humor is known to relieve individuals from psychological stress, it is necessary to understand how human beings adopted Internet Memes for coping up with the lockdown stress and stress-relieving mechanism during the lockdown period. In this paper, we have considered thirty popular memes and the increase in the number of their captions within the period (September 2017 to August 2020). An increase in Internet Meme activity since the lockdown period (March 2020) depicts an increase in online social behavior. We analyze the internet meme activity in social networks during the COVID-19 lockdown period using random forest, multi-layer perceptron, and instance-based learning algorithms followed by data visualization using line graph and Heat Map (8 & 15 clustered). We also compared the performance of the models using evaluation parameters like mean absolute error, root-mean-squared error & Kappa statistics and observed that random forest and instance-based learning algorithms perform better than multi-layer perceptrons. The result indicates that random forest and instance-based learning classifiers are having near perfect classification tendencies whereas multi-layer perceptrons showed around 97% classification accuracy.