Brain tumors significantly impact human health due to their complexity and the challenges in early detection and treatment. Accurate diagnosis is crucial for effective intervention, but existing methods often suffer from limitations in accuracy and efficiency. To address these challenges, this study presents a novel deep learning (DL) approach utilizing the EfficientNet family for enhanced brain tumor classification and detection. Leveraging a comprehensive dataset of 3064 T1‐weighted CE MRI images, our methodology incorporates advanced preprocessing and augmentation techniques to optimize model performance. The experiments demonstrate that EfficientNetB(07) achieved 99.14%, 98.76%, 99.07%, 99.69%, 99.07%, 98.76%, 98.76%, and 99.07% accuracy, respectively. The pinnacle of our research is the EfficientNetB3 model, which demonstrated exceptional performance with an accuracy rate of 99.69%. This performance surpasses many existing state‐of‐the‐art (SOTA) techniques, underscoring the efficacy of our approach. The precision of our high‐accuracy DL model promises to improve diagnostic reliability and speed in clinical settings, facilitating earlier and more effective treatment strategies. Our findings suggest significant potential for improving patient outcomes in brain tumor diagnosis.

Intrusion detection systems are essential for securing wireless sensor networks (WSNs) and Internet of Things (IoT) environments against various threats. This study presents a novel hybrid machine learning (ML) model that integrates KMeans-SMOTE (KMS) for data balancing and principal component analysis (PCA) for dimensionality reduction, evaluated using the WSN-DS and TON-IoT datasets. The model employs classifiers such as Decision Tree Classifier, Random Forest Classifier (RFC), and gradient boosting techniques like XGBoost (XGBC) to enhance detection accuracy and efficiency. The proposed hybrid (KMS + PCA + RFC) approach achieves remarkable performance, with an accuracy of 99.94% and an f1-score of 99.94% on the WSN-DS dataset. For the TON-IoT dataset, it achieves 99.97% accuracy and an f1-score of 99.97%, outperforming traditional SMOTE TomekLink and Generative Adversarial Network-based data balancing techniques. This hybrid approach addresses class imbalance and high-dimensionality challenges, providing scalable and robust intrusion detection. Complexity analysis reveals that the proposed model reduces training and prediction times, making it suitable for real-time applications.

An IoT healthcare system refers to the use of Internet of Things (IoT) devices and technologies in the healthcare industry. It involves the integration of various interconnected devices, sensors, and systems to collect, monitor, and transmit health-related data for medical purposes. Blockchain-assisted intrusion detection on IoT healthcare systems is an innovative approach to enhancing the security and privacy of sensitive medical data. By combining the decentralized and immutable nature of blockchain technology with intrusion detection systems (IDS), it is possible to create a more robust and trustworthy security framework for IoT healthcare systems. With this motivation, this study presents Blockchain Assisted IoT Healthcare System using Ant Lion Optimizer with Hybrid Deep Learning (BHS-ALOHDL) technique. The presented BHS-ALOHDL technique enables IoT devices in the healthcare sector to transmit medical data securely and detects intrusions in the system. To accomplish this, the BHS-ALOHDL technique performs ALO based feature subset selection (ALO-FSS) system to produce a series of feature vectors. The HDL model integrates convolutional neural network (CNN) features and long short-term memory (LSTM) model for intrusion detection. Lastly, the flower pollination algorithm (FPA) is exploited for the optimal hyperparameter tuning of the HDL approach, which results in an enhanced detection rate. The experimental outcome of the BHS-ALOHDL system was tested on two benchmark datasets and the outcomes indicate the promising performance of the BHS-ALOHDL technique over other models.

The early identification and treatment of various dermatological conditions depend on the detection of skin lesions. Due to advancements in computer-aided diagnosis and machine learning approaches, learning-based skin lesion analysis methods have attracted much interest recently. Employing the concept of transfer learning, this research proposes a deep convolutional neural network (CNN)-based multistage and multiclass framework to categorize seven types of skin lesions. In the first stage, a CNN model was developed to classify skin lesion images into two classes, namely benign and malignant. In the second stage, the model was then used with the transfer learning concept to further categorize benign lesions into five subcategories (melanocytic nevus, actinic keratosis, benign keratosis, dermatofibroma, and vascular) and malignant lesions into two subcategories (melanoma and basal cell carcinoma). The frozen weights of the CNN developed–trained with correlated images benefited the transfer learning using the same type of images for the subclassification of benign and malignant classes. The proposed multistage and multiclass technique improved the classification accuracy of the online ISIC2018 skin lesion dataset by up to 93.4% for benign and malignant class identification. Furthermore, a high accuracy of 96.2% was achieved for subclassification of both classes. Sensitivity, specificity, precision, and F1-score metrics further validated the effectiveness of the proposed multistage and multiclass framework. Compared to existing CNN models described in the literature, the proposed approach took less time to train and had a higher classification rate.

A cyber-communication infrastructure is required for physical components like control systems, sensors, actuators, and the environment to communicate and collaborate in real-time effectively. This is what cyber-physical systems (CPSs) provide. Deep Learning (DL) techniques have sped up and improved the progress of Intelligent Transportation Systems (ITS) in recent years, especially in problem domains traditionally addressed with analytical or statistical solutions. In addition to advancing driverless vehicle development to a new level, the improvements brought about by DL applications have led to better traffic organization and scheduling, increased security and safety on transit roads, reduced maintenance costs, and optimized performance for public transportation ride-sharing companies. The primary goal of this work is to offer a complete study and awareness of the uses of DL models on ITS, as well as to demonstrate the development in ITS research that has been owing to DL studies. This article briefly introduces the reader to various DL methods before thoroughly examining and describing how these methods are now being used in the transportation sector. Deep learning models are trained on this real-world traffic information to detect better and forecast the probability of crashes. This work aims to do just that by providing a multi-perspective assessment of deep learning-based techniques for traffic forecasting This paper offers a summary and taxonomy of current traffic forecast systems. We provide a compilation of the most cutting-edge methods currently used for traffic forecasting. In addition, we assess the efficacy of various approaches using a public, real-world dataset and provide an evaluation and analysis of our findings. These findings demonstrate that, related to state-of-the-art shallow models, a deep model is superior at traffic detection and achieves equivalent results in terms of traffic prediction. We suggest a new deep learning framework, an Attention-based hybrid Convolutional Neural network with Long Short Term Memory (LSTM) (AHCNLS), to perform real-time traffic prediction from a data mining approach to enhance driver and passenger safety. The suggested approach considers the spatial and temporal connection between GPS trajectories and contextual elements. Using a publicly available dataset, our proposed technique is evaluated and shown to have advantages over competing methods.

Plant disease detection is a critical challenge in agriculture, as undetected or poorly managed diseases can lead to significant yield losses, economic setbacks for farmers, and compromised food security. Therefore, accurate and efficient models for timely identification and mitigation are imperative to address these challenges and ensure sustainable agricultural practices. In this study, we introduce a deep multistacking integrated model for plant leaf disease detection that leverages fine-tuned transfer learning (TL) models, multistacking feature generation, and an ensemble XGBoost meta-classifier. Our approach involves specialized pipelines for image preprocessing, augmentation, and fine-tuning TL models, resulting in a robust hybrid model. The multistacking feature generation process aggregates the prediction probabilities from fine-tuned models, which are then used as input for the XGBoost classifier, enhancing both accuracy and efficiency. We evaluate the proposed model on three benchmark datasets: Tomato Disease Dataset (TDDS), Potato Pepper Dataset (PPDS), and Apple Grape Dataset (AGDS). Our experimental results demonstrate that the multistacking integrated model significantly outperforms traditional single-model approaches, achieving high accuracy scores of 99.78%, 99.86%, and 99.82% on TDDS, PPDS, and AGDS, respectively. These results underline the effectiveness of our approach, particularly in multilevel classification, where integrating features from multiple fine-tuned models allows for better generalization and higher precision. By leveraging multistacking techniques and an ensemble XGBoost classifier, we have advanced the state-of-the-art in plant disease detection. Moreover, the model shows superior computational efficiency, with lower build and prediction times compared to existing methods. These findings indicate that the integration of advanced ensemble techniques with fine-tuned models can enhance the generalization and robustness of plant disease detection systems. Our work provides a scalable and effective solution for improving plant disease detection, contributing to the broader application of deep learning (DL) in precision agriculture.