International Journal of Innovations in Science & Technology
Not a member yet
813 research outputs found
Sort by
Predictive Analytics for Smart Cities: Traffic Flow Forecasting Using Ensemble Algorithms
Full text linkTraffic flow prediction is crucial for smart transportation systems, as it plays a key role in improving traffic management and planning infrastructure. While many machine learning techniques have been used for this purpose, ensemble methods have proven to be especially effective because they enhance prediction accuracy by combining the strengths of multiple models. This paper offers a detailed overview of how ensemble methods are applied to traffic flow prediction. We start by exploring the basics of traffic flow prediction, including common data sources, types, and performance metrics. Then, we categorize ensemble methods into bagging, boosting, and hybrid approaches, reviewing important studies that show how these methods work, the datasets they use, and their performance results. Real-world examples and case studies are included to highlight the practical effectiveness of these methods in various traffic situations. Finally, we discuss the current challenges and suggest future research directions, aiming to provide a valuable resource for researchers and practitioners interested in improving traffic flow prediction with ensemble techniques
PsyRA – A Retrieval-Augmented Dialogue System for Mental Health Support
Full text linkMental health support continues to face numerous challenges, including limited access to care, persistent social stigma, and a shortage of trained mental health professionals. In response to these issues, this paper introduces PsyRA, an innovative AI-powered system designed to enhance psychological assessments through a specialized retrieval-augmented generation (RAG) approach. Unlike conventional chatbots that often fail to capture the nuanced context of patient interactions, PsyRA leverages domain-specific psychological knowledge to deliver more accurate and in-depth assessments. It draws from a carefully curated knowledge base that includes psychological research, diagnostic guidelines, therapy exercises, and intervention strategies to inform its responses and suggestions. PsyRA is equipped to understand patient narratives more clearly, provide evidence-based assessments by retrieving relevant psychological information, and offer personalized intervention recommendations tailored to individual needs. Early evaluations indicate that PsyRA is capable of detecting subtle emotional cues within patient conversations and responding in alignment with established psychological practices. The system demonstrates promising potential to broaden access to mental health support, assist professionals in the assessment process, and reduce the barriers that often prevent individuals from seeking treatment. This work contributes to the expanding field of AI-assisted mental health care by illustrating how retrieval-based models can enhance both the depth and quality of psychological assessments, offering improved emotional sensitivity and reliable, evidence-driven guidance
EEG Based BCI for Intelligent Wheelchair Control System Using Deep Learning
Full text linkThis research study presents the design of an Electroencephalography (EEG) based Brain Computer Interface (BCI) for intelligent wheelchair control to assist patients with mobility disorders. The concept of this research is to enable a direct communication link between the human brain and the machine without physical movement. This study used the BCI Competition IV 2a dataset, which contains EEG recordings of nine subjects performing four motor imagery (MI) tasks that were mapped to wheelchair navigation commands such as turning left, right, moving forward, and stopping. In this study, a deep learning architecture, TCFormer (Temporal Convolutional Transformer), was implemented to learn the spatial and temporal correlations between EEG channels. A lightweight Fusion Head module was added to enhance performance. It consisted of one-dimensional convolution and adaptive pooling operations for improved local temporal feature extraction. The proposed TCFormer-Fusion model achieved an overall classification accuracy of 75%, outperforming the baseline TCFormer model by 72%. Overall, this research study demonstrates that transformer-based models can learn complex EEG signal representations for motor imagery classification. The proposed model contributes toward developing an intelligent wheelchair control system that operates on brain signals, reducing external assistance. This work, with further optimization and real-time implementation, can contribute significantly to the assistive technology and human-computer interaction fields
Design and Validation of a Quantum Software Engineering Lifecycle Framework
Full text linkQuantum computing promises transformative computational capabilities; however, the absence of structured and standardized software engineering significantly limits its practical scalability. Quantum Software Engineering (QSE) remains fragmented, tool-centric, and largely experimental. This paper proposes a structured Quantum Software Engineering Lifecycle Framework that systematically integrates requirements engineering, hybrid design, quantum development, verification, deployment, and governance. To validate the proposed lifecycle, a two-stage evaluation was conducted. First, a Delphi-based expert validation involving 15 domain experts assessed clarity, feasibility, scalability, and hybrid applicability. Second, a simulation-based comparative analysis using Qiskit and PennyLane evaluated the lifecycle against ad-hoc development workflows across variational and search-based quantum algorithms. Results demonstrate a 24% reduction in development time, 15–18% improvement in execution fidelity, and significant gains in modularity, reusability, and tool interoperability. These findings confirm that adopting a structured lifecycle enables more reliable, scalable, and sustainable quantum software development, positioning QSE as a mature engineering discipline rather than purely experimental practice
CLFT: An Optimized Hybrid Cross-Layer Fusion Transformer for Accurate Fake Profile Detection on Social Media
Full text linkThe rapid increase of fake profiles on social media platforms has raised significant concerns regarding online authenticity, user trust, and digital security. Despite various efforts to combat this issue, existing detection methods often fall short due to the evolving nature of fake profiles and the noisy, high-dimensional data involved. This study proposes an optimized Hybrid Cross-Layer Fusion Transformer (CLFT) for detecting fake profiles by analyzing behavioral metadata. The CLFT architecture integrates multi-stage attention mechanisms, including Cross-Layer Fusion Attention (CLFA), Sparse–Dense Hybrid Attention (SDHA), and Temporal-Behavior Embedding Blocks (TBEB), to effectively capture both short- and long-term dependencies in user activities. The model hyperparameters were optimized using the Bayesian Optimization Hyperband (BOHB) framework. Experimental results on a real-world social media dataset show that the proposed model outperforms traditional machine learning techniques and previous Transformer-based models, achieving an accuracy of 99.10%, precision of 99.89%, recall of 99.55%, and an F1-score of 99.72%. Furthermore, the attention mechanisms enhance interpretability by emphasizing the most influential behavioral features, contributing to the model’s transparency and reliability. The findings highlight that Transformer-based models, especially the CLFT, provide a scalable and efficient solution for fake profile detection in noisy environments, with important implications for enhancing social media security. The study emphasizes the need for interpretability in automated detection systems, fostering trust and ensuring better user engagement and platform integrity
Over-Voltage Protection Circuit for Tripping & Switching of 220V Appliances Using Relay Module and Python-Based Algo
No full textMitigating the risk of appliance damage caused by overvoltage conditions is a critical aspect of ensuring electrical safety and reliability. This study presents a novel "Over Voltage Protection Circuit for Tripping and Switching 220V Appliances Using a Relay Module," designed to address this challenge through an automated disconnection mechanism. The proposed system enhances protection, operational stability, and the longevity of connected appliances, making it a valuable addition to electrical safety measures. The circuit utilizes a combination of electronic components including transistors, capacitors, resistors, LEDs, diodes, and a relay module. Through the integration of these elements, the circuit detects over-voltage situations, triggers the relay to disconnect the appliance, and indicates the status through LEDs. The adjustable potentiometer allows for customization of the overvoltage threshold, enhancing flexibility and adaptability to varying electrical environments. Test results demonstrated that the circuit reliably disconnected the load at the specified voltage threshold and reconnected it when conditions stabilized. The design successfully accounted for hysteresis in relay operation, minimizing unnecessary toggling and ensuring stable performance. This practical and cost-effective solution offers reliable overvoltage protection for diverse applications
Crypto Currency Compensation Model to Detect Optimal Channel of Internet of Things Through Blockchain
Full text linkThe ever-growing number of belongings of internet (IoT) devices in civilization creates a reliable, accessible, and safe infrastructure for processing the calculated data. One-point failures result from the prevalent IoT version\u27s use of an imperative cloud server approach. Because Blockchain uses a distributed community, IoT is integrated with Blockchain generation to avoid this. Consequently, this study has developed a fully autonomous and self-regulating learning system that can accurately operate channel time/spectral characteristics to communicate multi-user statistics. The future system is distinct in that it uses community metrics as its primary basis to recognize and adjust to increasing community density. Following the extraction of those capabilities, the projected protocol efficiently selects the appropriate channel for incoming nodes based on its interval features, recognizes and allocates the idle spectrum of nearby channels, and provides the optimal and appropriate channel utilization through an article called multilevel Gaussian radial and a multilayer non-linear assist vector machine (SVM) type model. The value consumption rate of the secure network and its functionalities is calculated in order to assess the performance of the proposed system. Future and conventional systems are compared. Associated to the prior model, the accuracy of the current model is 95.6%
Automated Objects Delivery System for Interior Locale using Line Following Robot with Optimized Security Parameters
Full text linkAutomated object delivery robots are increasingly sought for convenience, reliability, efficiency, supporting organizational productivity, elderly assistance, and reducing human error and labor costs in indoor delivery tasks. While various security measures have been implemented for the delivery robot’s safety, the design strategies used in existing studies do not suffice as they do not use biometric technology for unlocking the robot and real-time image tracking of robot thievery via mobile app. This research-based project aims to design and develop an object delivery system within small to medium-scale buildings using a robotic prototype controlled via an Android app. The robot navigates using a line-following technique with IR sensors, avoids static obstacles with an ultrasonic sensor, verifies the receiver with a fingerprint scanner, detects the destinations using an RFID module, and captures images of illicit attempts using an ESP32 camera module sending them in the app simultaneously. The designed prototype along with the Android app has undergone several feature tests with varying conditions. The results suggest that the system can securely carry a payload weighing 20 kg and is capable of navigating 10 km with a speed of 5 m/s depending upon the battery power. This project plans to tackle significant Sustainable Development Goals (SDGs) specifically, achieve Quality Education through SDG 4, Decent Work and Economic Growth in SDG 8, and Industry, Innovation, and Infrastructure in SDG 9
Load Balancing in Cloud Computing: A Proposed Novel Approach Based on Walrus Behavior
Full text linkThis research provides a comprehensive evaluation of load-balancing algorithms in cloud computing, classifying them into static, dynamic, and nature-inspired categories. Static algorithms, such as Round Robin and Min-Min, offer simplicity and efficiency in environments with stable workloads but struggle with adaptability to varying demands. Dynamic algorithms like Throttled Load Balancing and Least Connection are more flexible, adjusting to real-time server load changes and improving resource utilization, though they introduce higher overhead and computational costs. Nature-inspired algorithms, including Ant Colony Optimization and Particle Swarm Optimization, draw from biological processes to achieve high scalability, fault tolerance, and adaptability. A novel Walrus Optimization Algorithm (WaOA) is proposed, inspired by the social and migratory behaviors of walruses, to address challenges such as task bottlenecks and resource underutilization. MATLAB simulations reveal that WaOA outperforms traditional and nature-inspired methods in terms of scalability, response time, and resource optimization. The study concludes with suggestions for integrating machine learning, hybrid techniques, and real-world testing to further enhance WaOA’s effectiveness
Comparative Evaluation of Machine Learning and Deep Learning Models for Real Estate Price Prediction
Full text linkAccurate real estate price prediction plays a vital role in informed decision-making for investors, policymakers, and stakeholders. This study evaluates various machine learning and deep learning models for predicting real estate prices using the House Prices 2023 dataset which contains 168,000 entries of Pakistani property data. In our proposed methodology we performed data preprocessing and features engineering to standardize the data. We performed extensive experiments by using machine learning (ML) and deep learning (DL) models on our preprocessed data. The model’s performance was evaluated based on the R-squared (R²) score and Mean Squared Error (MSE) metrics. Based on the provided metrics, the Decision Tree achieved the highest performance with an R² of 0.9968 and an MSE of 0.0021, followed by Random Forest with an R² of 0.990 and MSE of 0.0007. Similarly, other ML models like Gradient Boosting and XG Boost also outperformed by achieving (R² 0.9959, MSE 0.0028 R² 0.9747, and MSE 0.0170) respectively. In contrast, models like AdaBoost, Neural Network, and Convolutional Neural Network (CNN) showed comparatively lower performance due to the nature of the data. The study emphasizes that ensemble-based models like Decision Trees and Random Forests are highly effective at identifying patterns in real estate prices. Additionally, applying optimization techniques improves the models ability to generalize and perform well on unseen data