Greenhouse gas Methane (CH$_4$) has 86 times more impact on global warming than carbon dioxide (CO$_2$). The emission of methane gas into the atmosphere is increasing due to the reliance on fossil-based resources in post-industrial energy consumption, along with the rise in food demand and the generation of organic waste that accompanies a growing human population. CH$_4$ acts as a vital pollutant in the air. The problem addressed in this study was to accurately estimate CH$_4$ emissions from functional urban areas. This study aims to predict CH$_4$ emissions using Time Series (TS) and Machine Learning (ML) models such as Autoregressive Integrated Moving Average (ARIMA), Seasonal ARIMA (SAR... More >

To address the challenge of effectively leveraging multi-source data for automated operational assessment of Unmanned Surface Vehicles (USVs) and utilizing digital technologies for monitoring and control, this paper proposes a data-driven state assessment method for surface unmanned systems and develops a digital twin system tailored for USVs. First, a dual-channel feature modeling mechanism is constructed by integrating physically interpretable statistical features with temporal convolutional features. Second, a complementary modeling strategy is adopted using CatBoost for static classification and GRU for dynamic modeling, while a Covariance Intersection (CI) fusion strategy is introduced... More >

Around the world, brain tumors are a major cause of human mortality. Accurate brain tumor detection is essential for effective treatment and improved patient outcomes. This study introduces the progressive transfer learning method, using VGG16 and MobileNet for the brain tumor identification and classification task. The outcome demonstrated the importance of the proposed models. The final accuracy of VGG16 and MobileNet on the test data was 98% and 87%, respectively, highlighting the superiority of VGG16 over the MobileNet framework. In addition, future work suggests advanced fine-tuning strategies, regularization techniques, and other methods to improve model performance for helping medical... More >

Adaptation learning is a data-driven technique that gives instructions based on the experiences made during data analysis. It plays an integral role in providing engineering solutions based on specific needs. Researchers have used the second-order statistics criterion for decades to conceptualize the optimality criteria using Shannon and Renyis information-theoretic measures. Some gaps have been identified in this research work, and useful findings have been proved with generalized information-theoretic measures of Renyis as Tsallis entropy of order $\alpha$ and Kapur entropy of order $\alpha$ and type $\beta$ using the Parzen-Rosenblatt window. This work explored the problem of constructing... More >

Early diagnosis of cardiac abnormalities depends on accurate classification of heart sounds, but centralized training methods run the danger of violating patient privacy. We thus propose a privacy-preserving and reliable heart sound abnormality detection system combining Blockchain Technology with Federated Learning (FL). Training is spread among seven clients, each simulating an independent data source, using a preprocessed dataset from the PhysioNet Challenge 2016 to enable distributed learning without sharing raw data. CNN-LSTM model using FedAvg achieved the best performance: 94\% accuracy, 0.90 precision, 0.96 recall, and an AUC of 0.98 among five deep learning architectures evaluated w... More >

The classification of Capsicum spp. varieties is often hindered by their morphological similarities, making accurate identification a challenging task. To address this issue, this study applies a hybrid computational approach that combines data dimensionality reduction techniques using Principal Component Analysis and Factor Analysis with various supervised Machine Learning algorithms. The dataset, which is unprecedented in the literature and was collected under controlled agricultural conditions, enables a robust evaluation of models including Logistic Regression, Support Vector Machine, K-Nearest Neighbors, Random Forest, Decision Tree, and Gradient Boosting. Model performance was assessed... More >

The healthcare sector has both opportunities and challenges as a result of the rapid expansion of unstructured clinical text data in electronic health records (EHRs). Physician notes, reports from radiologists, and summaries of discharge are examples of narrative medical documents from which relevant and actionable information can be extracted using clinical text analytics driven by Natural Language Processing (NLP). Named entity recognition, conceptual normalization, relation extraction, and temporal reasoning are just a few of the core methods and approaches in clinical natural language processing that are thoroughly covered in this paper. It covers cutting-edge deep learning models like B... More >

Deep learning is a great success primarily because it encodes large amounts of data and manipulates billions of model parameters. Despite this, it is challenging to deploy these cumbersome deep models on devices with limited resources, such as mobile phones and embedded devices, due to the high computational complexity and the amount of storage required. Various techniques are available to compress and accelerate models for this purpose. Knowledge distillation is a novel technique for model compression and acceleration, which involves learning a small student model from a large teacher model. Then, that student network is fine-tuned on any downstream task to be applicable for resource-constr... More >

The paper carries out a comparative study that is based on the use of credibility theory to examine pentagonal and hexagonal fuzzy membership functions of machine learning systems. These fuzzy memberships can be used to manage the uncertainty and imprecision of a data driven-model which allows better decision-making in the case of vague or incomplete information. The credibility theory is used to determine quantitatively the reliability of the inferences obtained through each function. Both the membership functions are modelled, incorporated in machine learning framework and tested on randomly generated as well as application specific datasets. The results obtained indicate that the performa... More >

This paper investigates the core principles of quantum computation, providing an in-depth understanding of quantum phenomena and illustrating how these principles form the scientific foundation of the field. The pivotal physical concepts, such as properties of subatomic particles, including electrons and photons, as well as their mathematical description through linear algebra are examined. It focuses on the qubit, the quantum analogue of a classical bit, featuring properties like superposition, entanglement, and wave function collapse, which redefine the traditional concept of information processing. The mathematical structures that underlie quantum system modelling—vector spaces, tensor... More >

Deep learning has substantially enhanced facial emotion recognition, an essential element of human--computer interaction. This study evaluates the performance of multiple architectures, including a custom CNN, VGG-16, ResNet-50, and a hybrid CNN-LSTM framework, across FER2013 and CK+ datasets. Preprocessing steps involved grayscale conversion, image resizing, and pixel normalization. Experimental results show that ResNet-50 achieved the highest accuracy on FER2013 (76.85%), while the hybrid CNN-LSTM model attained superior performance on CK+ (92.30%). Performance metrics such as precision, recall, and F1-score were used for evaluation. Findings highlight the trade-off between computational e... More >

The Autism Spectrum Disorder (ASD) diagnosis and detection in its initial stages is a more complex issue in the face of the wide-ranging, diverse nature and causes. Subsequent literature inclined towards a possible correlation of gut microbiome with ASD, and its disclosure presents a more promising attribute for imminent discovery conduits. The dataset on gut microbiome associated with ASD focuses specifically on the microbial compositions obtained through 16S rRNA sequencing. This study presents a novel method that integrates Artificial Intelligence employing various Machine Learning (ML) robust classifiers such that Support Vector Machines (SVM), Random Forest, k-Nearest Neighbors (KNN), L... More >

In the age of increasing machine-mediated communication, the ability to detect emotional nuances in speech has become a critical competency for intelligent systems. This paper presents a robust Speech Emotion Recognition (SER) framework that integrates a hybrid deep learning architecture with a real-time web-based inference interface. Utilizing the RAVDESS dataset, the proposed pipeline encompasses comprehensive preprocessing, data augmentation techniques, and feature extraction based on Mel-Frequency Cepstral Coefficients (MFCCs), Chroma features, and Mel-spectrograms. A comparative experiment was run against a standard machine learning classifier such as K-Nearest Neighbors (KNN), Support... More >

Air pollution has emerged as a critical challenge, directly affecting human health, urban sustainability, and climate systems. Traditional air-quality index (AQI) prediction models often struggle to provide timely alerts because they are not very sensitive to changes over time and are hard to understand. This paper proposes a Neuro-Inspired Alert System for Air Quality Prediction (NAS--AQP) that incorporates an ensemble learning approach using voting regression to enhance input quality, followed by classification through a Spiking Neural Network (SNN). The system is designed such that it captures the temporal and nonlinear relationships between air pollutants such as Nitrogen Dioxide ($NO_2$... More >

The capacity of robots to produce captions for images independently is a big step forward in the field of artificial intelligence and language understanding. This paper looks at an advanced picture captioning system that uses deep learning techniques, notably convolutional neural networks (CNNs) and recurrent neural networks (RNNs), to provide contextually appropriate and meaningful descriptions of visual content. The suggested technique extracts features using the DenseNet201 model, which allows for a more thorough and hierarchical comprehension of picture components. These collected characteristics are subsequently processed by a long short-term memory (LSTM) network, a specific RNN variat... More >

Neutrosophic sets play a significant role for handling indeterminacy. In this paper, we introduce a novel fuzzy non-linear regression model to find the minimum spread of neutrosophic fuzzy sets. Kuhn-Tucker's necessary conditions are employed to estimate the parameters for non-linear regression models, which can be applied to any data set. The resulting hybrid model possesses the ability to minimise the spread of uncertainty in a much better fashion than the existing non-linear regression contenders which rely on KKT- based model. The hybrid approach reduces the maximum spread by 22.09% and improves prediction accuracy, as shown by a 22.23% reduction in RMSE. The study’s findings highligh... More >

To meet latency constraints, fog computing takes computational assets to the network edge. Blockchain and reinforcement learning are increasingly being integrated into the Industrial Internet of Things (IIoT) to enhance security and efficiency. This study introduces a Reinforcement Learning-based Resource Scheduling Approach for Blockchain Networks in IIoT. Unlike previous studies, which mainly focus on either blockchain security or resource allocation, our approach integrates reinforcement learning for dynamic resource scheduling, improving efficiency while minimizing latency. The methodology is illustrated through a flowchart. Simulation results validate the effectiveness in multiple scena... More >

This research proposes an advanced system for landmine detection combining the internet of things and reinforcement learning, which seeks to resolve issues in conventional methods that misidentify more than 30% of detections, have slow reaction times, and are not suited for different environments. Others like metallic detectors and sniffer dogs also pose greater danger for wrong threat identification, more so due to slothful attempts. The system proposed in this study is novel in that it customizes metal detection by integrating a sensor into military boots, thus permitting constant scanning without the use of hands. A metaplastic Machine Learning model improves detection accuracy. It was fo... More >

Traditional centralized machine learning approaches for IoT botnet detection pose significant privacy risks, as they require transmitting sensitive device data to a central server. This study presents a privacy-preserving Federated Learning (FL) approach that employs Federated Averaging (FedAvg) to detect prevalent botnet attacks, such as Mirai and Gafgyt, while ensuring that raw data remain on local IoT devices. Using the N-BaIoT dataset, which contains real-world benign and malicious traffic, we evaluated both the IID and non-IID data distributions to assess the effects of decentralized training. Our approach achieved 97.5% accuracy in IID and 95.2% in highly skewed non-IID scenarios, clos... More >

Machine intelligence has evolved from being a purely theoretical idea into a fundamental element of contemporary technology, transforming industries and influencing society on a broad scale. This editorial delves into its historical development, recent advancements, and prospective future directions. It highlights the dynamic interaction between technological progress, innovative algorithms, and the ethical challenges that shape the field, offering a thorough and insightful overview. More >