A Memory Efficient Adversarial Attention Tree-Structured Deep Learning Model for Classification

Nirmala; Jayanthi

doi:https://doi.org/10.54216/JISIoT.160205

Fusion: Practice and Applications | 2025

A Novel Computer Vision-Based Approach to Mitigating Fall Risks in the Elderly through Spatial-Channel Decoupled Downsampling in YOLOv10

Elderly health has always been a matter of concern for the medical doctors and researchers to come up with advanced recovery techniques. With the rise in population of elderly people and mostly residing alone at home in solitude has motivated many researchers to work on remedial measures for the biggest safety risk faced by them which is elderly fall prevention and mitigating thereby causes of injuries. In this paper, an intelligent deep learning and computer vision based elderly fall recognition system is designed which utilizes advanced spatial-channel decoupled downsampling in You Only Look Once version 10 (YOLOv10), pytorch, darknet and cascaded CNN technologies for the fall detection. The results after testing manifest that the accuracy of the proposed system to recognize and detect the elderly fall is quite assuring, the values of accuracy and mean Average Precision (mAP50) coming out to be 92.46% and 94.1% respectively after the model validation. Moreover, the system displays a real time performance as it can process approximately 45 frames of images per second that realizes a real-time identification of elderly fall patterns. As compared to previous models, the proposed model is much more efficient and has shown promising results.

groups

Ajay Singh mail -

Alok Katiyar mail

link https://doi.org/10.54216/FPA.190219

Volume & Issue

Vol. Volume 19 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

A Novel CNN Model for Fruit Leaf Disease Detection: A Lightweight Solution for Grapes, Figs, and Oranges

Plant diseases are considered a real threat to food security due to the losses incurred by individuals and countries. Early detection is one of the real solutions that can help reduce the size of these losses, but early detection is still bleeding. This study presents the development of a Convolutional Neural Network (CNN) model for classification with a new architecture and optimal performance suitable for real-time applications for the detection of fruit diseases (figs, oranges, grapes). The developed CNN model balanced accuracy and FLOPs using Squeeze-Excitation (SE) and adaptive-average pool layers. After implementing new data developed from Iraqi farms, the CNN model achieved optimal performance compared to the most famous models such as VGG16, ResNet, EfficientNet, and AlexNet.

groups

Dalya Anwar mail

link https://doi.org/10.54216/FPA.190220

Volume & Issue

Vol. Volume 19 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

HI2NN: Heuristic Intelligence towards Enhancing Rainfall Prediction with Improved Artificial Neural Networks

Predicting rainfall proves critical for businesses to organize their water resources, make agricultural choices, and prevent disasters. Therefore, proposed model presents a novel approach, namely Heuristic Intelligence towards Enhancing Rainfall Prediction with Artificial Neural Networks (HI2NN) to enhance rainfall prediction by designing heuristic Intelligence combined with Improved Artificial Neural Networks (IANNs). The proposed HI2NN framework leverages heuristic optimization techniques to fine-tune ANN parameters to improve prediction accuracy. Prediction accuracy is computed through our designed custom accuracy metric. The methodology uses historical weather information to extract complex non-linear patterns, which neural models generate from the designed big dataset. The accuracy level of rainfall predictions using our methodology achieves 92%, which demonstrates superior performance than traditional approaches that include random forest and decision tree and XGBoost models. The new forecasting systems develop higher reliability through collaborative efforts between heuristic algorithms and neural networks as described in this research work targeting challenging meteorological forecasts.

groups

Sachin Subhashrao Patil mail -

Sonali Ridhorkar mail

link https://doi.org/10.54216/FPA.190221

Volume & Issue

Vol. Volume 19 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

A Transfer Learning-Driven Framework for Enhanced Software Development Effort Estimation Using Optimized Hybrid Deep Learning Model

Precise assessment of software development effort (SDE) is essential for efficient project planning and resource distribution. Conventional methods frequently encounter difficulties in generalizing across different project areas because of disparate data attributes. This research presents an innovative approach that combines transfer learning with hybrid deep learning models to tackle these difficulties. The platform utilizes pre-trained Random Forest and LSTM models, enhanced using Jaya optimization, to improve prediction accuracy and adapt effectively to new datasets. Transfer learning is utilized to extract reusable patterns and features from source domains, facilitating effortless adaption to target domains with minimum retraining. Extensive experiments on various benchmark datasets illustrate the proposed framework's enhanced performance regarding accuracy, scalability, and robustness relative to leading techniques. This study emphasizes the capability of transfer learning to transform SDE estimates, providing a scalable and domain-adaptive approach for intricate software projects.

groups

Badana Mahesh mail -

Mandava Kranthi Kiran mail

link https://doi.org/10.54216/FPA.190222

Volume & Issue

Vol. Volume 19 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Disease Prediction Using Machine Learning Approaches Considering Bio-Medical Signal Analysis: A Survey

In medical diagnosis and prognosis, symptoms provided by patients play a critical role in identifying diseases. Machine learning offers a powerful approach to analyzing and predicting illnesses based on these symptoms. In particular, classification algorithms are widely used to analyze input data and predict disease outcomes. A key factor in effective classification is the selection of relevant attributes, which directly affects the accuracy of the prediction. This research emphasizes the importance of proper feature extraction techniques in the context of disease prediction using biomedical signal analysis. Effective analysis requires both the extraction of critical features and the elimination of irrelevant data. The aim of this study is to explore existing approaches to disease prediction based on biomedical signal analysis. We focus on feature extraction from pre-processed data, which aids in distinguishing between different biomedical signals recorded by medical devices. Our objective is to identify biomedical cues that differentiate various health conditions. Examples of such signals include electroencephalogram (EEG), electrocardiogram (ECG), and electrogastrogram (EGG). Understanding how these signals differ between healthy and diseased states is crucial for accurate disease prediction. This research investigates diseases such as heart disease, kidney failure, and lung infections, considering how variations in biomedical signals can be used to predict the likelihood of severe illness. We continue to seek advancements in predicting and mitigating future health risks

groups

K. Satyanarayana Murthy mail -

Suribabu Korada mail

link https://doi.org/10.54216/FPA.190223

Volume & Issue

Vol. Volume 19 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Fusion of Real-Time Traffic and Environmental Sensor Data with Machine Learning for Optimizing Smart City Operations

The complex developing nature of urban infrastructure necessitates intelligent solutions for optimizing smart city operations. Based on this research paper, a multi-modal fusion framework that integrates real-time traffic and environmental sensor data with advanced machine learning algorithms to enhance decision-making for urban traffic management and pollution control is proposed. A hybrid AI model is proposed, with a combination of CNNs for the estimation of image-based traffic density, LSTM networks for the time-series environmental prediction, and RL for adaptive control of traffic signals. The system proposed integrates sensor data in real-time from cameras, GPS, LiDAR, and nodes for environmental monitoring to create an optimized control strategy. The model has been deployed on edge computing devices, such as Raspberry Pi, to enable the real-time processing and reduce the latency. Security layer based on block chain for data integrity protection and tamper proofing within smart city networks. The suggested system shows high improvements in congestion reduction, better accuracy in air pollution forecasting, and energy efficiency in urban management. It will be validated using simulation with SUMO and MATLAB and real-world sensor data that the sensor fusion approach outperforms the conventional fixed-rule strategies of traffic management. This work allows for cost-effective, large-scale smart city deployment that would reduce traffic delay and urban air pollution while securing data and being computationally efficient. The low-latency decision-making approach with edge-AI makes it fit for real-time urban governance. Unlike traditional models that process either traffic or environmental data in silos, the work presented herein integrates multi-source sensor data with edge computing and blockchain security for a unified AI-driven fusion approach, thus building a robust framework for next-generation smart city intelligence.

groups

Harish Reddy Gantla mail -

Sunil Kr Pandey mail -

Shailaja Mantha mail -

Priya Goyal mail -

Asmath Jabeen mail -

Shameem Fatima mail -

Udit Mamodiya mail

link https://doi.org/10.54216/FPA.190224

Volume & Issue

Vol. Volume 19 / Iss. Issue 2

Details open_in_new

Journal of Intelligent Systems and Internet of Things | 2025

Resources Management Consıderıng Envıronmental Condıtıons in Educatıonal Instıtutıons Based on IoT

One of the most significant issues affecting the majority of countries in the world today is resource conservation. Water is the most vital component for all life, hence protecting it is crucial. Optimal use of water maintains its sustainability and leads to energy savings. Educational institutions are considered among the largest institutions that use water because of the presence of large numbers of students and employees. This research concerned resource management in educational institutions taking into account environmental conditions based on Internet of Things (IoT). The results illustrated that the designed monitoring system for moisture content has the ability to enhance water sustainability by using the optimal water content. A significant efficiency of the proposed monitoring system in controlling the water level was achieved. The maximum error between the monitoring system reading and the actual reading was 2% and 2.44% for moisture content and water level, respectively. The results showed the sensor's high sensitivity to rainfall and the ability of the proposed monitoring system to save water that exceeds the need of soil

groups

M. E. ElAlami mail -

M. M. Ghoniem mail -

Asmaa E. El-Maghraby mail

link https://doi.org/10.54216/JISIoT.160201

Volume & Issue

Vol. Volume 16 / Iss. Issue 2

Details open_in_new

Journal of Intelligent Systems and Internet of Things | 2025

An Effective IoT based Vein Recognition Using Convolutional Neural Networks and Soft Computing Techniques for Dorsal Vein Pattern Analysis

In this research, we provide a CNN-based system that can reliably identify the dorsal veins of the hand. In order to get better results on different picture quality datasets, the suggested model makes use of refined variants of the pre-trained VGG Net-16 and VGG Net-19 designs. We use the BOSPHORUS dataset, which provides medium-quality photos, in addition to two self-constructed datasets that provide good- and low-quality images. By using state-of-the-art augmenting image methods, streamlined pre-processing procedures, and meticulously designed CNN designs, the fine-tuned VGG Net-16 model achieves superior performance in comparison to all other models. Using ROI pictures with a resolution of 224×224 pixels, a multi-class technique is employed for arranging the vein patterns. Improving data quality during training makes the approach more broad, which helps prevent over fitting. On every dataset, the proposed method achieves better results than standard ML models like K-NN and SVM, and the experimental outcomes demonstrate significant improvements in accuracy. The modifying process led to a considerable decrease in the equal error rates (EER) when compared to benchmark methods. The structure enhances efficiency in computing with GPU-accelerated studying. It was built with the help of Python extensions like as OpenCV, Keras, and TensorFlow. Results from extensive testing of the proposed method show an accuracy of 99.98%, a precision of 98.98%, and a recall of 98.8%. From what we can see, the technique is both adaptable and dependable; making it well suited for use in practical biometrics vein recognition applications.

groups

Krishna Bhimaavarapu mail -

Bylapudi Rama Devi mail -

Chandra Bhushan Mahato mail -

Lakshmi Chandrakanth Kasireddy mail -

M. Vadivukarassi mail -

P. Sivaraman mail

link https://doi.org/10.54216/JISIoT.160203

Volume & Issue

Vol. Volume 16 / Iss. Issue 2

Details open_in_new

Journal of Intelligent Systems and Internet of Things | 2025

Transforming Public Health with AI and Big Data Deep Learning for COVID-19 Detection in Medical Imaging

For public health systems worldwide, the COVID-19 epidemic has presented hitherto unheard-of difficulties. Rapid and accurate virus detection is essential for successful treatment and containment. This paper explores the transformative potential of Artificial Intelligence (AI) and Big Data in public health, focusing on applying deep learning techniques for COVID-19 detection in medical imaging. We discuss the integration of AI-driven solutions in healthcare, the role of big data in enhancing diagnostic accuracy, and the implications for future public health strategies. The COVID-19 pandemic started in Dec 2019 and has wreaked havoc on our lives ever since. One such youngest addition to the coronavirus family has claimed the lives of almost half the world's population. With the introduction of constantly evolving forms of this infection, locating the infection early on would still be essential. Even though the PCR test is the best and most utilized approach for identification, non-contact procedures such as chest radiography and CT scans have always been recommended. In this context, artificial intelligence is integral to the early and precise diagnosis of COVID-19 via lung image processing. The primary aim of this study is to evaluate and contrast multiple deep learning improved strategies for detecting COVID-19 in CT and X-Ray medical images. We employed four strong CNN methods for the COVID-19 images of the binary classification challenge: ResNet152, VGG16, ResNet50, and DenseNet121. The suggested Attention-based ResNet framework is created to choose the appropriate architecture and training settings for models automatically. In the diagnosis of COVID-19 utilizing CT-scan images, the accuracy and F1-score are over 96 percent. In addition, transfer-learning methods were used to address the lack of information and shorten the training time. Enhanced VGG16 deep transfer learning design was used to accomplish multi-class categorization of X-ray imaging tasks. Enhanced VGG16 was shown to have 99 percent accuracy in detecting X-ray imaging from three classes: Normal, COVID-19, and Pneumonia. The algorithms' accuracy and validity were tested on well-known public datasets of X-ray and CT scans. For COVID-19 diagnosis, the presented approaches outperform previous methods in the literature. In the fight against COVID-19, we believe our research will aid virologists and radiologists in making better and faster diagnoses.

groups

Md Jabed Hussain mail -

Awakash Mishra mail

link https://doi.org/10.54216/JISIoT.160204

Volume & Issue

Vol. Volume 16 / Iss. Issue 2

Details open_in_new

Journal of Intelligent Systems and Internet of Things | 2025

A Memory Efficient Adversarial Attention Tree-Structured Deep Learning Model for Classification

The representational and learning power of tree-based deep-learning (DL) classification models makes them a popular choice for dimensional sentiment analysis (DSA). One variant, Tree-structured Convolutional neural network with long short-term Memory (TCL) stands out among many others for its ability to handle uncertainties and unexpected changes in input data while still producing promising Valence-Arousal (VA) predictions for text or image classes. However, the high memory complexity of this model becomes a challenge when dealing with large image/text datasets. To address this issue, this manuscript introduces a Lightweight Adversarial Attention TCL (LAATCL) model for DSA. The proposed model includes a clustering layer in conjunction with the ATCL to decrease memory complexity and enhance performance through reliable sample selection. This model comprises multi-convolution with a clustering layer that utilizes Group-Sparse Non-negative Matrix Factorization (GSNMF) for clustering highly correlated samples. By learning informative and discriminative latent variables across labels, GSNMF helps identify and select samples closest to the cluster centroid for input to the LSTM network, resulting in reduced memory complexity and improved accuracy. The LATCL model outperformed traditional models in experiments conducted on the SST and CIFAR-10 datasets, with accuracies of 93.57% and 95.25%, respectively, demonstrating its usefulness.

groups

Nirmala Veluswamy mail -

Jayanthi Boopathy mail

link https://doi.org/10.54216/JISIoT.160205

Volume & Issue

Vol. Volume 16 / Iss. Issue 2

Details open_in_new