| Paper ID |
IJIFR/V13/E8/058
|
| Author |
Dr.J.PRIYADHARSHINI, DEPARTMENT OF COMMERCE(CORPORATE SECRETARYSHIP),SONA COLLEGE OF ARTS AND SCIENCE, SALEM-05
Dr. EZHILVANI C M, DEPARTMENT OF COMMERCE(CORPORATE SECRETARYSHIP),SONA COLLEGE OF ARTS AND SCIENCE, SALEM -05
Dr.R.RADHA, DEPARTMENT OF COMMERCE(CORPORATE SECRETARYSHIP), SONA COLLEGE OF ARTS AND SCIENCE, SALEM-05
|
| Paper Title |
FINANCIAL INCLUSION INITIATIVES IN INDIA
|
| Subject Category |
COMMERCE
|
| Abstract |
Financial Inclusion and Education are two important elements in the Reserve Bank of India's developmental role. Towards this, it has created critical volume of literature and has uploaded on its website in 13 languages for banks and other stakeholders to download and use. The aim of this initiative is to create awareness about financial products and services, good financial practices, going digital and consumer protection.
Financial inclusion refers to ensuring that individuals and businesses can access essential financial products and services, such as savings accounts, loans, insurance, and payment services, at affordable prices. Financial inclusion aims to eliminate the barriers that prevent people from participating in the financial sector and using its services to improve their lives. It is also known as inclusive finance.
In India, financial inclusion has been pivotal for poverty alleviation, bridging the income divide, and supporting national economic objectives. By fostering financial literacy, promoting digital banking, and improving access to formal banking channels, financial inclusion aims to achieve sustainable economic and social equity.
|
| Keyword |
Financial Inclusion, Banking, Digital Payment, Financial Inclusion Initiatives, Challenges, National Strategy, Entrepreneurship.
|
|
| Paper ID |
IJIFR/V13/E8/057
|
| Author |
Dr. Savita Panwar, Deputy Director, IGNOU Regional Centre, Chandigarh
|
| Paper Title |
Comparative analysis of enrollment trends of IGNOU in Chandigarh and Dehradun Regional Centres
|
| Subject Category |
Education
|
| Abstract |
In the paper an attempt has been made to do a comparative analysis of session-wise enrolment trend of IGNOU Regional Centre Chandigarh (RCC) and IGNOU Regional Centre Dehradun (RCD) since 2010 to 2025. The paper also focuses on the various student support services provided by both the Regional centres. The comparative analysis of various factors such as use of ICT, social media platform to reach out to the learners for the enhancement of Gross Enrolment Ratio (GER) as per the NEP2020. The difference in the socio-demographic factors of both the regions and its effect of the session-wise enrolment of both the Regional centres. The effect of COVID-19 on the enrolment and the delivery of online and offline student support services by both the RCs and their Learner Support Centres (LSCs). The various strategies used by both the RCs for promotion and publicity of IGNOU programmes on offer and the new programmes being launched from time to time to the prospective learners in the geo-physical inaccessible areas and the marginalized section of the society. For the purpose of the present study the data has been collected from primary and secondary sources. The analysis of the paper is presented in the coherent frame of the study.
|
| Keyword |
Gross Enrolment Ratio (GER), NEP2020, ICT, Publicity and Promotion, Regional Centres, Learners Support Centres
|
|
| Paper ID |
IJIFR/V13/E8/056
|
| Author |
Jasmine, Dept. of Computer Science and Engineering, SRM Institute of Science and Technology ,Kattankulathur, India
Arihant Jain, Dept. of Computer Science and Engineering SRM Institute of Science and Technology ,Kattankulathur, India
P. Rajasekaran, Dept. of Computer Science and Engineering SRM Institute of Science and Technology ,Kattankulathur, India
J. D. Dorathi Jayaseeli, Dept. of Computer Science and Engineering SRM Institute of Science and Technology ,Kattankulathur, India
|
| Paper Title |
ASL-Edu: An AI-Driven Platform for American Sign Language Education
|
| Subject Category |
Computer Science and Engineering
|
| Abstract |
Disabling hearing loss affects more than 430 million individuals globally, yet the bulk of educational media continues to be produced in spoken form with little meaningful accom- modation for signing communities. We introduce ASL-Edu, a web-based system designed to bridge this gap by transforming spoken instructional video into American Sign Language (ASL) narration and constructing structured sign-language lessons from arbitrary topics on demand. Four distinct AI subsystems work in concert: an OpenAI Whisper speech recognizer attaining an 8.3% word error rate on educational recordings; a gloss-matching retrieval engine operating over the 2,000-entry WLASL clip library with 68.4% lexical coverage on classroom transcripts; a Google Gemini 1.5 Pro module that composes lessons whose sentence structure respects ASL grammatical ordering; and a Stable Diffusion XL generator that produces illustrative hand- configuration imagery tailored to each lesson concept. Delivered as a React 18 single-page application fronting a FastAPI service with SQLite persistence, the full pipeline resolves a two-minute educational video into an ASL-narrated presentation within
3.2 seconds on average. Thirty hearing-impaired participants evaluated the system, awarding a mean score of 4.2 out of 5 across sign accuracy and lesson clarity dimensions. The results suggest that unifying these heterogeneous AI capabilities behind a single, accessible interface meaningfully reduces the barrier to quality sign-language education relative to the fragmented landscape of existing single-purpose tools
|
| Keyword |
American Sign Language, automatic speech recognition, sign language education, generative AI, Whisper, WLASL, Stable Diffusion, Gemini, multimodal learning, accessibility
|
|
| Paper ID |
IJIFR/V13/E8/055
|
| Author |
Yosshmi Nair, Dept. of Computer Science and Engineering, SRM Institute of Science and Technology ,Kattankulathur, India
Vashistha Mehta, Dept. of Computer Science and Engineering SRM Institute of Science and Technology ,Kattankulathur, India
P. Rajasekaran, Dept. of Computer Science and Engineering SRM Institute of Science and Technology ,Kattankulathur, India
J. D. Dorathi Jayaseeli, Dept. of Computer Science and Engineering SRM Institute of Science and Technology ,Kattankulathur, India
|
| Paper Title |
CyberShield AI: A Unified Platform for Detecting Deepfakes, AI-Generated Imagery, Phishing URLs, Credit-Card Fraud, and Document Tampering
|
| Subject Category |
Computer Science and Engineering
|
| Abstract |
The shift to remote work, online banking, and syn- thetic media has pushed digital threats out of narrow categories and into overlapping, fast-moving territory. A single phishing email today may carry a forged invoice, a deepfaked voice clip, and a link that fingerprints the victim’s browser. Most defensive tools, however, were built for one threat at a time. This paper describes CyberShield AI, a unified web platform that runs five separate detection pipelines under one interface: a deepfake video detector based on InceptionV3 with temporal smoothing, an AI- generated image detector built on a frequency-aware ResNet- 18 (the FIRE design) that fuses RGB channels with FFT mid- frequency features, a phishing URL classifier built on XGBoost over a sixteen-feature lexical and host vector, a credit-card fraud detector trained with Random Forest on a heavily imbalanced transaction set, and a document-tampering module that combines Error Level Analysis with a convolutional network. The platform is delivered as a React 18 single-page application backed by a FastAPI service, with model artifacts hosted on Hugging Face Spaces and media handled through Cloudinary. Across our test sets, the five modules reach 88.77%, 92.5%, 96.85%, 97.86%, and 77.2% accuracy respectively, and serve predictions in 0.4 s to 6.5 s depending on modality. We argue that consolidating these checks into one workflow lowers the cognitive cost of verification for non- technical users and shortens the response time for analysts who would otherwise switch between four or five tools.
|
| Keyword |
deepfake detection, AI-generated image de- tection, phishing URLs, credit card fraud, document forgery, multimodal machine learning, web security, FastAPI, React
|
|
| Paper ID |
IJIFR/V13/E8/054
|
| Author |
Sakshi Sharan, Law Student, School of Law, Christ (Deemed to be University), Delhi NCR, Uttar Pradesh, India
|
| Paper Title |
Gender Justice at Work: Evaluating India Labour Law Protections and the Implementation Gap for Women in the Formal Retail Sector
|
| Subject Category |
Law
|
| Abstract |
The increasing women's participation in the workforce represents a major change in the social and economic conditions of India. However, women still face difficulties and challenges in achieving equality at work, even with the existence of a strong set of labour laws and progressive policy initiatives and laws protecting them.
This research paper critically examines the regulatory framework that governs women’s employment in the formal retail sector in India and analyses the protective framework and the implementation challenges that lead to inefficiency. The paper uses a mixed-method approach, which includes doctrinal analysis of legal provisions that are specifically attuned to women's needs, which includes welfare and health provisions under the Factories Act(1948), the Maternity Benefit Act (2017), the POSH Act (2013), social security laws and the new consolidated labour codes, and empirical insights that are gathered through semi-structured interviews with women workers of the organised retail sector. This methodology aims to provide a detailed view of the gap between law in books and law in action and provide comprehensive recommendations.
The paper argues that while India's laws look good on paper, there are still many issues concerning their implementation, such as a lack of knowledge of the rights, poor enforcement, unsafe working hours, informal deductions, and limited facilities at workplaces. The recommendations are inclusive of the employer compliance, targeted inspections being performed, and workplace awareness raising, as well as compulsory social security enrollment and seating arrangements with enforcement, and retail chains being monitored.
|
| Keyword |
Women workforce participation; Indian labour law; gender justice; maternity benefits; occupational safety; labour codes; POSH; implementation gap
|
|
| Paper ID |
IJIFR/V13/E8/053
|
| Author |
RESMI N C, Vels institute of Science Technology and Advanced Studies( VISTAS )
Dr.R. Jeyanthi, Vels Institute of Science Technology and Advanced Studies
|
| Paper Title |
Technological Pedagogical Content Knowledge (TPACK) and Teacher Self-Efficacy: A Comprehensive Review of Related Literature
|
| Subject Category |
Education
|
| Abstract |
The integration of technology in education has necessitated a redefinition of teacher competencies, emphasizing the importance of Technological Pedagogical Content Knowledge (TPACK) and teacher self-efficacy. This paper presents a comprehensive review of literature examining the relationship between TPACK and self-efficacy and their collective influence on teaching effectiveness. Drawing on empirical and theoretical studies, the review identifies key themes, including the positive association between TPACK and self-efficacy, the mediating role of self-efficacy, the influence of professional development, and inconsistencies in the alignment between perceived competence and actual knowledge. The review also highlights contextual and demographic factors affecting these constructs. The study concludes by identifying research gaps and proposing directions for future research, particularly in developing integrated models and examining mediating variables.
|
| Keyword |
TPACK, teacher self-efficacy, pedagogical content knowledge, technology integration, literature review
|
|
| Paper ID |
IJIFR/V13/E8/052
|
| Author |
REXEN JACOB R, Assstat Professor & Research Scholar (Zamorins Guruvayurappan College Kozhikode) PG Department Of Sociology Govt. Knm College Kanjiramkulam, Kerala
|
| Paper Title |
INSTITUTINAL CARE FOR ELDERLY IN KERALA: A STUDY AMONG OLDER PERSONS IN OLD AGE HOME ALAPPUZHA, KERALA
|
| Subject Category |
Sociology
|
| Abstract |
Today the world has been experiencing the proportion of old age citizens (above 60 years) in India constitutes 8.6% of the total population (60 million). About 25% of this population currently resides in old age homes across India due to various reasons. The reasons may be social, economic, cultural and familial grounds. The present living conditions of elderly in old age homes are not commendable. But these institutions are providing a cherishing memory and address the vulnerabilities of elderly regarding economy and health conditions. It has been found from research that the setting of old age homes and the employees influence the welfare of the residents and their health care. Despite various services provided, researchers have found that service gaps exist among staff and residents of old age homes s a witnessing issue. In the present scenario, it is the need to address and understand the living standards of the older persons in old age homes, and to study the various services offered by them and how the residents spend their lives there in order to better their conditions of living with considering their economic and health vulnerabilities. The main objectives of the present study include (i) to identify the socio-economic background and roe of old age care homes for elderly (ii) To examine the services of care homes for the health vulnerabilities of elderly (iii) to analyse the institutional support for handling the economic vulnerabilities of elderly in Kerala. The study was based on mixed research methodology and structured interview schedule and guide was used to collect the data from the respondents of Govt Old Age Home Alappuzha, Kerala. The analysis of the data shows that problems the inmates face is: physical disabilities, heath problems, emotional problems, lack f support from children and family members. The source of fund is from donations and they do not engage in any promotional activities.
|
| Keyword |
Older persons, care home, old age homes, institutional support, economic vulnerability, health vulnerability
|
|
| Paper ID |
IJIFR/V13/E8/051
|
| Author |
Gajula Sai Rohith, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V. Vijayalakshmi, Associate Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
SmartRetail AI: An Ensemble Machine Learning Framework for Demand Forecasting and Inventory Intelligence in Retail Supply Chain Management
|
| Subject Category |
Computer Science
|
| Abstract |
Contemporary retail operations are confronted with escalating complexity in demand forecasting and inventory management, driven by non-linear demand variability, promotional amplification, seasonal fluctuations, and the competitive imperatives of omnichannel commerce. Conventional forecasting paradigms—predicated upon moving averages, exponential smoothing, and linear extrapolation—demonstrably fail to capture the multivariate, non-linear interactions governing retail demand dynamics. This paper introduces SmartRetail AI, a comprehensive, end-to-end retail intelligence platform that integrates ensemble machine learning with classical inventory optimisation theory within a unified data architecture. The system is constructed upon a structured synthetic dataset encompassing 4,250 daily transaction records across five representative Fast-Moving Consumer Goods and e-commerce product categories, spanning 850 operational days. The proposed forecasting engine deploys product-level Random Forest Regressor models—each trained independently to capture category-specific demand dynamics—to generate thirty-day forward demand forecasts. Engineered temporal features including day-of-week indicators, monthly seasonality encodings, weekend binary flags, and promotional activity markers constitute the input feature space. The inventory optimisation module applies the classical safety stock formulation at a 95% service level target (Z = 1.65), computing product-specific reorder points as a function of historical demand variability and a three-day supply lead time. Analytical outputs are persisted in a normalised MySQL relational schema comprising three tables—historical sales, demand forecasts, and inventory metrics—and rendered through an executive intelligence dashboard. Comparative evaluation against moving average baselines demonstrates Mean Absolute Error reductions of 23.4% to 47.8% across product categories, with Root Mean Square Error improvements ranging from 19.7% to 44.1%. The platform achieves a Mean Absolute Percentage Error of 12.7% for high-volume staples and 24.3% for low-velocity electronics, establishing its operational viability across diverse retail demand profiles. These results validate the proposed framework as a practically deployable, analytically rigorous alternative to legacy forecasting methodologies.
|
| Keyword |
Demand Forecasting; Random Forest Regressor; Inventory Optimisation; Safety Stock; Retail Intelligence; Ensemble Machine Learning; Supply Chain Management
|
|
| Paper ID |
IJIFR/V13/E8/050
|
| Author |
Gurramkoda Charan Kumar, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
S. Usharani, Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V.Vijayalakshmi, Associate Professor, Department of MCA, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
ShopMind 360: An Integrated Multi-Algorithm Analytics Engine for E-Commerce Customer Behavioral Segmentation, Predictive Scoring, and Association-Rule-Based Personalization
|
| Subject Category |
Computer Engineering
|
| Abstract |
Contemporary e-commerce platforms generate voluminous transactional and behavioral data streams that, absent a structured analytical framework, remain commercially inert. ShopMind 360 is proposed as a comprehensive, end-to-end behavioral analytics and personalization engine that synthesizes four complementary machine learning and data mining methodologies within a single, automated pipeline. The system operationalizes Recency-Frequency-Monetary (RFM) analysis for multidimensional customer value quantification, K-Means clustering for behavioral segmentation into four actionable customer archetypes (Champions, Loyal Customers, At-Risk, and Hibernating), Random Forest ensemble classification for continuous purchase-probability scoring, and the Apriori algorithm for market-basket association rule mining to generate confidence-ranked product recommendations. The analytical pipeline ingests a synthetic yet realistically parameterized dataset comprising 500 customer profiles, 50 product records, 2,000 transactional orders, and 5,000 browsing interaction logs, structured within a multi-sheet Excel workbook and subsequently persisted to a normalized MySQL relational database. All computed analytical results are surfaced through an interactive Power BI dashboard, providing business stakeholders with filterable, real-time-refreshable visualizations of customer segments, revenue contributions, purchase probability distributions, and product affinity rules. Experimental evaluation demonstrates that the four-segment K-Means model achieves stable cluster centroids with well-separated RFM profiles, the Random Forest classifier attains high discriminative accuracy in identifying high-value customer segments, and Apriori mining yields statistically significant association rules with lift values substantially exceeding unity. The system architecture adheres to modular design principles, enabling independent maintenance and extensibility of each analytical component without pipeline restructuring. ShopMind 360 establishes a replicable, open-source blueprint for data-driven customer engagement in small-to-medium-scale e-commerce operations.
|
| Keyword |
Recency-Frequency-Monetary Analysis; K-Means Clustering; Random Forest Classification; Apriori Association Rule Mining; Customer Behavioral Segmentation
|
|
| Paper ID |
IJIFR/V13/E8/049
|
| Author |
Avadutha Prathibha, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
S. Usharani, Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
An Intelligent AI-Driven Clinical Decision Support System for Symptom-Based Disease Prediction Using Random Forest Ensemble Learning
|
| Subject Category |
Computer Engineering
|
| Abstract |
The increasing complexity of healthcare systems and the global shortage of medical professionals have amplified the need for intelligent clinical decision support systems capable of assisting in early disease diagnosis. This paper presents an AI-driven medical diagnosis support system that leverages machine learning techniques to predict diseases based on patient-reported symptoms. The proposed system utilizes a Random Forest Classifier trained on a high-dimensional dataset comprising over 130 symptoms mapped to more than 40 disease categories. The model employs multi-hot encoding for symptom representation and generates probabilistic predictions with associated confidence scores.
The system is implemented as a web-based application using the Django framework, enabling role-based interaction for both patients and doctors. Patients can input symptoms through an intuitive interface, while doctors gain access to aggregated diagnostic insights and patient history. The machine learning pipeline integrates feature importance extraction, allowing visualization of symptom influence using Chart.js, thereby enhancing interpretability.
Experimental results demonstrate high classification accuracy and robust performance across diverse symptom combinations. The system significantly reduces diagnostic latency and provides preliminary clinical insights, particularly in resource-constrained environments. The integration of AI with web-based healthcare systems highlights the potential for scalable, accessible, and efficient diagnostic assistance tools.
|
| Keyword |
Artificial Intelligence; Clinical Decision Support; Random Forest; Disease Prediction; Machine Learning
|
|
| Paper ID |
IJIFR/V13/E8/048
|
| Author |
C. Nawaz Basha, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
S.Usharani, Professor, Department of Computer Applications, Viswam Engineering College,Madanapalli,Andhra Pradesh
|
| Paper Title |
An Integrated Random Forest-Based Framework for Insurance Claim Amount Regression and Fraud Risk Classification in Imbalanced Datasets
|
| Subject Category |
Computer Science
|
| Abstract |
The insurance industry confronts two analytically critical and financially consequential challenges: accurate prediction of claim settlement amounts and timely detection of fraudulent claims. Conventional approaches — rule-based heuristics, logistic regression scorecards, and manual adjuster assessments — are demonstrably inadequate for capturing the nonlinear, high-dimensional interactions that characterise modern insurance claim data. This paper presents ClaimSmart AI, a comprehensive, modular, end-to-end machine learning pipeline that addresses both challenges within a unified analytical framework. The system operates on a synthetically generated dataset of 15,000 insurance claim records encompassing 19 attributes spanning policyholder demographics, policy characteristics, vehicle parameters, claim specifics, and behavioural indicators. A dual-model architecture employs a Random Forest Regressor (150 estimators) for claim amount prediction and a Random Forest Classifier (150 estimators, balanced class weights) for binary fraud risk detection, both trained on a stratified 80/20 holdout split with StandardScaler feature normalisation and LabelEncoder categorical transformation. The regression model achieves a Mean Absolute Error below INR 15,000 and an R-squared coefficient of determination exceeding 0.70, while the classification model delivers accuracy above 0.80, fraud-class recall exceeding 0.74, and F1-Score above 0.76, surpassing logistic regression and rule-based baselines on equivalent evaluation protocols. Prediction outputs are enriched with four derived business metrics — predicted claim amount, claim variance, fraud risk probability, and a three-tier fraud risk category — and persisted to a MySQL relational database for direct consumption by Power BI and enterprise analytics platforms. Eight publication-quality visualisation charts provide comprehensive analytical coverage from fraud distribution and regional heatmaps to actual-versus-predicted scatter analysis. A mysqldump-format SQL export module ensures enterprise portability and regulatory archival compliance. The complete pipeline executes through a single orchestration script, establishing ClaimSmart AI as both a rigorous academic contribution and a practical template for production insurance analytics deployment.
|
| Keyword |
Random Forest; Insurance Fraud Detection; Claim Amount Regression; Imbalanced Classification; Business Intelligence Integration
|
|
| Paper ID |
IJIFR/V13/E8/047
|
| Author |
G Mallikarjuna Reddy, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V.Vijayalakshmi, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
S.Usharani, Professor, Department of Computer Applications, Viswam Engineering College,Madanapalli,Andhra Pradesh
|
| Paper Title |
AI-Based Online Voting System: Integrating Isolation Forest Anomaly Detection, NLP Sentiment Analysis, and TF-IDF Candidate Matching in a Secure Django Electoral Platform
|
| Subject Category |
Computer Engineering
|
| Abstract |
Online voting systems represent one of the most demanding intersections of cybersecurity, civic participation, and artificial intelligence, requiring simultaneous guarantees of voter authentication, vote uniqueness, ballot integrity, fraud detection, and transparent auditability within a framework accessible to non-technical citizens. Existing e-voting platforms either rely on proprietary cryptographic protocols with limited AI-driven fraud detection, or prioritise accessibility while sacrificing the robust integrity mechanisms demanded by high-stakes elections. This paper presents an AI-Based Online Voting System that addresses these co-requirements through four integrated AI-driven components embedded within a production-quality Django 5.x web application.
The system architecture implements: (1) an Isolation Forest-based anomaly detection engine that computes a real-time risk score for every vote submission from three behavioural features — hour of submission, minute of submission, and IP address vote frequency — flagging suspicious events within the same HTTP request cycle without perceptible latency; (2) a TextBlob NLP sentiment analysis module that computes a manifesto polarity score (range -10 to +10) for each candidate biography using lazy computation triggered on first election page view; (3) a TF-IDF and cosine similarity AI Candidate Matcher that ranks candidates by alignment with voter-submitted preference text using Scikit-learn's TfidfVectorizer with English stop-word removal; and (4) a SHA-256 cryptographic vote hashing mechanism that generates a tamper-evident fingerprint for each Vote record at save time.
A multi-role architecture distinguishes Registered Voters, Election Administrators, and Audit Officers, with role enforcement through Django's @login_required and @user_passes_test decorators. Vote uniqueness is guaranteed through dual-layer enforcement: application-level duplicate checking and database-level unique_together constraints on the (voter, election) pair. Empirical evaluation on a 500-vote simulation dataset demonstrates a fraud detection precision of 91.2%, a false positive rate of 4.3%, a candidate matching accuracy of 89.7% against expert preference alignment, and a mean end-to-end vote processing latency of 87 milliseconds on consumer CPU hardware. The system is implemented entirely on open-source technologies, providing a reproducible reference architecture for AI-enhanced civic technology.
|
| Keyword |
Online voting system; Isolation Forest; anomaly detection; TF-IDF candidate matching; NLP sentiment analysis; SHA-256 vote integrity; Django; electoral security
|
|
| Paper ID |
IJIFR/V13/E8/046
|
| Author |
Madde Sanghavi, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
B. Shireesha, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
NutriGen AI – Personalized Nutrition Recommendation Engine Using Machine Learning
|
| Subject Category |
Computer Science
|
| Abstract |
The global burden of lifestyle-related diseases including obesity, type-2 diabetes, and cardiovascular conditions has established personalized nutrition as a healthcare necessity. Conventional dietary guidance, grounded in population-averaged standards such as Recommended Dietary Allowances, fails to account for individual metabolic variation. NutriGen AI addresses this limitation through a machine learning-powered personalized nutrition recommendation engine. The system employs a Random Forest Regressor trained on one thousand synthetic user profiles to predict Total Daily Energy Expenditure (TDEE), a hybrid filtering recommendation architecture combining K-Nearest Neighbors content-based filtering with Truncated Singular Value Decomposition collaborative filtering to produce meal suggestions, and goal-oriented macro-nutrient distribution logic for weight loss, maintenance, and muscle gain objectives. Delivered through a Flask RESTful backend and a glassmorphism-styled HTML5/CSS3/JavaScript frontend with Chart.js visualizations, the system democratizes access to personalized nutritional guidance using entirely open-source technologies. Experimental evaluation demonstrates effective TDEE estimation and nutritionally aligned meal recommendations across diverse user profiles.
|
| Keyword |
Personalized Nutrition, Machine Learning, Random Forest Regressor, Hybrid Recommendation System, KNN, Collaborative Filtering, TDEE Estimation, Flask, Health Informatics
|
|
| Paper ID |
IJIFR/V13/E8/045
|
| Author |
Mutra Rekha, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
B. Shireesha, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
QualityPredict AI – Manufacturing Quality Score Prediction System Using Ensemble Machine Learning
|
| Subject Category |
Computer Engineering
|
| Abstract |
Smart manufacturing (Industry 4.0) demands real-time predictive intelligence to eliminate reactive quality management. QualityPredict AI is a comprehensive, end-to-end machine learning platform for smart factories that predicts continuous product quality scores and classifies manufacturing defects by analyzing production telemetry including temperature, pressure, mechanical vibration, machine rotational speed, ambient humidity, and machine age. The system is trained on a synthetic dataset of ten thousand manufacturing records generated to replicate real-world industrial conditions. Three ensemble regression algorithms — Random Forest Regressor, LightGBM Regressor, and Gradient Boosting Regressor — are comparatively evaluated for quality score prediction, while a Random Forest Classifier provides binary defect detection with calibrated probability scores. Feature importance analysis reveals mechanical vibration as the dominant quality predictor (˜63% variance explained), followed by machine age (˜22%) and operating temperature (˜8%). The best regression model achieves an R² score exceeding 0.88 on a held-out 20% test split, and the defect classifier achieves accuracy above 0.90. Four professional analytical visualizations communicate findings to production managers and quality engineers. The complete pipeline from data generation through model training, visualization, and live prediction simulation is implemented in modular Python, with all model artifacts serialized via Joblib for production deployment.
|
| Keyword |
Manufacturing Quality Prediction, Industry 4.0, Random Forest, LightGBM, Gradient Boosting, Defect Classification, Feature Importance, Smart Manufacturing, Statistical Process Control, Predictive Quality Management
|
|
| Paper ID |
IJIFR/V13/E8/044
|
| Author |
BIJI B, SREE NARAYANA COLLEGE,SIVAGIRI,VARKALA
|
| Paper Title |
Empowering Small Businesses: A Deep Dive into Kerala’s MSME Support Schemes
|
| Subject Category |
COMMERCE
|
| Abstract |
Micro, Small, and Medium Enterprises (MSMEs) play a pivotal role in economic development by generating employment, fostering innovation, and promoting inclusive growth. In the Indian state of Kerala, MSMEs are a cornerstone of the industrial ecosystem, supported by a range of government initiatives aimed at financial assistance, skill development, and ease of doing business. This study explores the effectiveness and scope of Kerala’s MSME support schemes, including capital subsidies, credit facilitation programs, and digital governance initiatives. It examines how these schemes contribute to entrepreneurial growth, particularly among marginalized groups such as women and first-time entrepreneurs. Despite a well-structured policy framework, challenges such as procedural complexity, limited awareness, and accessibility barriers persist. The paper highlights the need for improved implementation strategies, better outreach, and continuous policy refinement to maximize the impact of these schemes. Overall, Kerala’s MSME model offers valuable insights into building a sustainable and inclusive small business ecosystem.
|
| Keyword |
MSMEs, Entrepreneurship, Government Schemes, Economic Development, Small Business Support, Kerala
|
|
| Paper ID |
IJIFR/V13/E8/043
|
| Author |
Kondakavli Vani, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V. Vijayalakshmi, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
SafeHelmet Vision AI: Industrial Safety Helmet Detection System
|
| Subject Category |
Computer Science
|
| Abstract |
Construction sites are among the most hazardous working environments, where head injuries caused by falling objects are a leading cause of fatalities. Ensuring the consistent use of safety helmets is therefore critical, yet traditional monitoring methods relying on manual supervision are inefficient and prone to human error. This paper presents SafeHelmet Vision AI, an intelligent deep learning-based system designed to automatically detect helmet usage among workers in construction environments. The proposed system utilizes the YOLOv8 (You Only Look Once version 8) object detection algorithm to identify and localize safety helmets and human workers in real-time images. The model is trained using transfer learning on a curated dataset of construction site images, enabling high accuracy even under challenging conditions such as varying lighting, occlusions, and diverse worker appearances. The system is deployed through a user-friendly web interface built using the Streamlit framework, allowing users to upload images and instantly receive detection results with annotated bounding boxes and safety compliance summaries. The system achieves a high detection performance with a mean Average Precision (mAP50) of 97.8%, precision of 95.2%, and recall of 94.9%, demonstrating its effectiveness for real-world industrial safety monitoring. Additionally, an automated alert mechanism identifies safety violations when workers are detected without helmets, enabling proactive intervention.
|
| Keyword |
Safety Helmet Detection; YOLOv8; Computer Vision; Deep Learning; Industrial Safety Monitoring
|
|
| Paper ID |
IJIFR/V13/E8/042
|
| Author |
K. Aparna, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V. Vijayalakshmi, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
INTELLIGENT COLLEGE PLACEMENT MANAGEMENT PLATFORM
|
| Subject Category |
Computer Science
|
| Abstract |
Campus placement plays a vital role in connecting engineering students with employers, but many institutions still rely on manual methods such as spreadsheets and emails, leading to inefficiencies. The Intelligent College Placement Management Platform is a full-stack web application developed using the Django framework to automate and streamline the entire placement process.The system supports three user roles: students, recruiters, and placement officers. Students can create profiles, upload resumes, apply for jobs, and track application status. Recruiters can post job openings, review applications, and issue offer letters. Placement officers manage the system through a centralized dashboard that provides real-time insights into placement activities.A key feature of the platform is the simulated AI-based matching engine, which uses natural language processing techniques to compare student skills with job requirements and generate a match score. This improves decision-making for both students and recruiters. The system also enforces placement policies such as the One-Student-One-Offer rule.Built using Python, Django, SQLite, and Bootstrap, the platform ensures efficient, transparent, and scalable placement management, improving coordination and reducing manual effort in campus recruitment processes
|
| Keyword |
Campus Placement System, Django Web Application, Placement Management, AI Matching Engine, Natural Language Processing (NLP), Resume Screening, Job Recommendation, Student Recruitment, Skill Matching, Full Stack Development
|
|
| Paper ID |
IJIFR/V13/E8/041
|
| Author |
Kadiri Bhuvaneswari, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V. Vijayalakshmi, Associate Professor & Research Supervisor, PG Department of Commerce VTM NSS College, Dhanuvachapuram, Thiruvananthapuram.
|
| Paper Title |
ResumeMatch Pro AI: A Transformer-Based Semantic Matching Engine for Intelligent Recruitment Support
|
| Subject Category |
Computer Engineering
|
| Abstract |
The contemporary recruitment landscape is characterised by an exponentially growing volume of applications that overwhelm keyword-based Applicant Tracking Systems (ATS), which treat language as a bag of isolated tokens and systematically fail to recognise semantically equivalent competency descriptions. This paper presents ResumeMatch Pro AI, a full-stack intelligent recruitment support system that addresses these representational inadequacies by deploying Sentence-BERT (SBERT), specifically the all-MiniLM-L6-v2 pre-trained transformer model, to encode both resume and job description documents into 384-dimensional dense semantic vector representations. Cosine similarity computed between these embeddings yields a holistic semantic match score that is robust to paraphrase, synonymy, and terminological variation — failure modes that fundamentally undermine conventional lexical matching. The system further employs the spaCy natural language processing library in conjunction with a PhraseMatcher-based skill extractor to perform fine-grained skill gap analysis, enumerating precisely which required competencies are present in the candidate profile and which are absent, thereby transforming an abstract score into an actionable decision-support artefact. The architecture follows a clean two-tier client-server separation: a FastAPI backend exposes RESTful endpoints for single-candidate matching and multi-candidate ranking, whilst a React/Vite frontend renders match results through circular gauge visualisations, colour-coded skill tags, and animated result panels designed for non-technical recruiters. Experimental evaluation using representative professional domain test cases demonstrates that the SBERT-based approach correctly resolves synonym ambiguities — crediting a candidate describing experience in 'statistical learning' against a role requiring 'machine learning' — where keyword systems assign a zero-overlap score. The system achieves single-request response times of one to three seconds on CPU-only infrastructure, confirming practical deployability. The proposed framework demonstrates that the combination of transformer-based holistic semantic embeddings with explicit rule-based skill extraction yields a recruitment tool that is simultaneously more accurate, more transparent, and more actionable than the lexical-matching status quo.
|
| Keyword |
Sentence-BERT; Semantic Resume Matching; Recruitment Automation; Cosine Similarity; Skill Gap Analysis
|
|
| Paper ID |
IJIFR/V13/E8/040
|
| Author |
Jeripiti Reddy Prasad, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V.Vijayalakshmi, Associate Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
SecurePay Shield: An Ensemble Machine Learning Architecture for Real-Time Fraud Detection and Probabilistic Risk Scoring in the Indian Digital Payment Ecosystem
|
| Subject Category |
Computer Science
|
| Abstract |
The exponential growth of digital financial transactions in India has precipitated a commensurate escalation in sophisticated payment fraud, necessitating intelligent, adaptive detection systems capable of operating at scale. This paper presents SecurePay Shield, a comprehensive, end-to-end machine learning pipeline engineered for real-time identification of fraudulent transactions within the Indian digital payment ecosystem. The proposed system addresses three principal challenges endemic to fraud detection: severe class imbalance, high-dimensional heterogeneous feature spaces, and the requirement for probabilistic, interpretable risk assessments amenable to regulatory scrutiny. The architecture employs an ensemble learning strategy integrating three complementary algorithms: a Random Forest Classifier (200 estimators), a Gradient Boosting Classifier (150 estimators), and an Isolation Forest anomaly detection model. A domain-specific feature engineering pipeline transforms 23 raw transaction attributes into a 35-dimensional feature space, computing composite risk indicators including the Risk_Composite score, IP_Risk_Score, and Velocity_Score, which collectively emerge as the most discriminative predictors of fraudulent activity. Class imbalance is mitigated through the application of the Synthetic Minority Over-sampling Technique (SMOTE), yielding a balanced training corpus of 22,560 instances. The Random Forest model, selected as the production deployment candidate, achieves an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 1.0000 and an F1-Score of 1.0000 on the held-out test partition, with five-fold stratified cross-validation yielding a mean F1 of 0.9978 (±0.0021), confirming model robustness and generalizability. Predictions and analytical artifacts are persisted in a structured MySQL database comprising five normalized tables, and operational insights are surfaced through a four-page Microsoft Power BI dashboard supporting real-time fraud monitoring. The system is demonstrated on a synthetic dataset of 15,000 Indian financial transactions and evaluated against 2,000 prospectively generated records, achieving a holistic prediction corpus of 17,000 transactions with a four-tier risk stratification (Critical, High, Medium, Low).
|
| Keyword |
Fraud Detection; Ensemble Learning; SMOTE; Random Forest; Digital Payment Security; Risk Scoring; Feature Engineering
|
|
| Paper ID |
IJIFR/V13/E8/039
|
| Author |
Koppala Jagadeesh, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V.Vijayalakshmi, Associate Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr.S.Usharani, Professor, Department of Computer Applications, Viswam Engineering College,Madanapalli,Andhra Pradesh
|
| Paper Title |
AI-Driven Insurance Claim Processing System: Automating Validation, Workflow Orchestration, and Decision Support Using Django and Rule-Based Intelligence
|
| Subject Category |
Computer Engineering
|
| Abstract |
The insurance industry processes millions of claims annually through predominantly manual, paper-based workflows that impose substantial administrative overhead, systemic processing delays, and significant vulnerability to fraudulent submissions. These inefficiencies translate directly into customer dissatisfaction, escalating operational costs, and regulatory compliance risk. Despite incremental digitisation efforts in recent decades, the majority of mid-tier and small-scale insurance operators continue to rely on manual adjudication processes that lack systematic validation mechanisms, consistent decision frameworks, and real-time transparency for policyholders.
This paper presents an AI-Driven Insurance Claim Processing System, a comprehensive web-based platform developed using the Django 4.x framework, Python 3.10, and SQLite — engineered to automate and orchestrate the complete lifecycle of insurance claim management. The proposed system implements a three-tier Model-View-Template (MVT) architecture encompassing a responsive presentation layer, a rule-based application logic engine, and a relational data persistence layer. Seven functionally decomposed modules address user authentication, role-based access control, policy management, automated claim validation, administrative adjudication, real-time status notification, and database lifecycle management.
The core contribution of the system is a deterministic automated validation engine that evaluates each submitted claim against two critical conditions — policy coverage limit adherence and policy temporal validity — eliminating ineligible claims at the point of submission without human intervention. Validated claims are routed to an administrative dashboard providing centralised oversight, statistical summaries, and structured approval workflows. Empirical evaluation on a simulated dataset of 500 claims demonstrates a claim processing time reduction of 74.3% relative to a manual baseline, a validation accuracy of 99.6%, and a false positive rejection rate of 0.4%. The system's modular Django architecture ensures extensibility toward future integration of machine learning-based fraud detection, OCR-driven document processing, and cloud-scale PostgreSQL deployment.
|
| Keyword |
Insurance claim processing; Django MVT architecture; automated validation; rule-based classification; role-based access control; digital workflow automation; claim adjudication
|
|
| Paper ID |
IJIFR/V13/E8/038
|
| Author |
Kondakavali Vani, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V.Vijayalakshmi, Associate Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr.S.Usharani, Professor, Department of Computer Applications, Viswam Engineering College,Madanapalli,Andhra Pradesh
|
| Paper Title |
SafeHelmet Vision AI: Real-Time Construction Site Safety Helmet Detection Using YOLOv8 Transfer Learning and Streamlit Deployment
|
| Subject Category |
Computer Engineering
|
| Abstract |
Construction sites consistently rank among the most hazardous occupational environments worldwide, with head injuries from falling or flying objects identified as a primary contributor to construction fatalities in every major market. Despite the universal regulatory mandate for safety helmet usage, non-compliance remains pervasive owing to the practical impossibility of maintaining continuous manual supervision across large, complex sites. Traditional automated monitoring approaches based on conventional computer vision techniques have demonstrated insufficient accuracy for reliable deployment in the visually complex conditions typical of active construction environments, while commercial AI-based platforms impose subscription costs prohibitive to small and medium contractors.
This paper presents SafeHelmet Vision AI, a deep learning-based industrial safety monitoring system designed to automate helmet compliance detection at construction sites and related industrial workplaces. The proposed system employs a YOLOv8n (nano) object detection model trained via transfer learning from COCO-pretrained weights on a domain-specific dataset of 4,200 annotated construction site images encompassing 9,800 helmet and 8,200 worker bounding box instances. Training was conducted for 100 epochs with AdamW optimisation (lr0 = 0.001), comprehensive data augmentation including mosaic, HSV perturbation, random flip, rotation (±10°), and scale variation (±50%), yielding a validation mean Average Precision at IoU = 0.50 (mAP50) of 97.8%, a precision of 95.2%, and a recall of 94.9%.
The trained model is integrated into a Streamlit web application that accepts uploaded construction site images in JPG, PNG, or BMP formats and returns annotated detection results with bounding boxes, class labels, and confidence scores within a mean inference latency of 165 milliseconds on standard CPU hardware. An automated safety compliance assessment engine evaluates helmet-to-person count ratios and generates colour-coded violation alerts. Comparative evaluation demonstrates a 27.2 percentage-point precision advantage and a 33.9 percentage-point recall advantage over a traditional Haar cascade baseline. The complete system requires no client-side installation and is deployable to Streamlit Cloud from a GitHub repository with a single configuration step, making enterprise-grade safety monitoring accessible to safety personnel without specialised technical training.
|
| Keyword |
Safety helmet detection; YOLOv8; construction site safety; personal protective equipment; transfer learning; real-time object detection; Streamlit deployment
|
|
| Paper ID |
IJIFR/V13/E8/037
|
| Author |
Pudu Bhargava, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
S. Manjunath Reddy, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
EMPLOYEE PERFORMANCE AND ATTRITION PREDICTION USING MACHINE LEARNING AND GENERATIVE AI
|
| Subject Category |
Computer Science
|
| Abstract |
Employee attrition is a major challenge for organizations, leading to productivity loss, higher recruitment costs, and disruption in operations. This paper presents an Employee Performance and Attrition Prediction System that combines HR management with machine learning and generative AI. The system is built using Django with SQLite as the database and uses Python libraries such as Scikit-learn for prediction, Openpyxl for data export, and Google Gemini API for generating explanations. It stores structured employee data including performance, attendance, and evaluations for analysis. A classification model predicts attrition risk based on factors like performance rating, projects completed, and salary. To improve interpretability, a generative AI module explains predictions in simple HR-friendly language. Overall, the system improves data management, enhances prediction accuracy, and provides an easy-to-understand, scalable solution for workforce analysis and decision-making.
|
| Keyword |
Seismic Forecasting; LSTM; ARIMA; Time Series Analysis; Disaster Management
|
|
| Paper ID |
IJIFR/V13/E8/036
|
| Author |
Peddapalli Manikanta, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
S. Manjunath Reddy, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
LEXIGUARD AI: MULTILINGUAL LEGAL INTELLIGENCE & RISK ANALYTICS ENGINE
|
| Subject Category |
Computer Engineering
|
| Abstract |
The increasing complexity of legal documents has created a significant barrier for individuals and organizations lacking access to professional legal expertise. This paper presents LexiGuard AI, a multilingual legal intelligence and risk analytics system designed to automate contract analysis using Natural Language Processing (NLP) techniques. The system processes legal documents in multiple formats, detects clause types using a keyword-based classification engine, and computes a comprehensive risk score based on weighted legal significance. A key contribution of this work is the integration of explainable AI through a SHAP-inspired mechanism that highlights word-level importance within legal clauses, enabling users to understand the reasoning behind risk assessments. The system also incorporates extractive summarization, multilingual support, and a conversational AI assistant for interactive legal guidance. The proposed architecture is implemented as a full-stack web application using FastAPI and modern frontend technologies. Experimental results demonstrate that the system effectively identifies critical clauses such as liability, indemnity, and intellectual property, while providing transparent and actionable insights. LexiGuard AI contributes toward democratizing access to legal intelligence by offering an accessible, scalable, and interpretable solution for contract risk evaluation.
|
| Keyword |
Legal Document Analysis; Natural Language Processing; Risk Assessment; Explainable AI; Contract
|
|
| Paper ID |
IJIFR/V13/E8/035
|
| Author |
Patnam Jayasree, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Manjunath Reddy, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr. S. Usharani, Professor & Head, Department of MCA ,Viswam Engineering College, Andhra Pradesh, India
|
| Paper Title |
SEISMO PREDICT AI: SEISMIC ACTIVITY FORECASTING SYSTEM USING LSTM AND ARIMA
|
| Subject Category |
Computer Science
|
| Abstract |
Earthquakes are among the most devastating natural disasters due to their sudden occurrence and the severe damage they cause to human life and infrastructure. Traditional seismic monitoring systems primarily focus on real-time detection and post-event analysis, offering limited capability for forecasting future seismic activity. This paper presents SeismoPredict AI, an intelligent seismic activity forecasting system that leverages both deep learning and statistical approaches to analyze historical earthquake data and predict future trends. The proposed system integrates Long Short-Term Memory (LSTM) networks to capture complex non-linear temporal dependencies and Autoregressive Integrated Moving Average (ARIMA) models to provide stable and interpretable time-series forecasts.The system is designed with a user-friendly interface using Streamlit, enabling users to upload datasets, visualize seismic patterns, train predictive models, and generate forecasts without requiring advanced technical expertise. Additionally, the system incorporates automated risk classification and alert generation mechanisms to support early warning and disaster preparedness. Experimental analysis demonstrates that the hybrid LSTM–ARIMA approach improves prediction reliability and trend consistency compared to individual models. The proposed system serves as an effective decision-support tool for researchers, policymakers, and disaster management authorities by providing meaningful insights into seismic activity trends. Although precise earthquake prediction remains inherently uncertain, the system contributes to proactive risk assessment and enhances preparedness strategies through data-driven forecasting
|
| Keyword |
Seismic Forecasting; LSTM; ARIMA; Time Series Analysis; Disaster Management
|
|
| Paper ID |
IJIFR/V13/E8/034
|
| Author |
Aluganti Vishnu Priya, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
M.Gowthami, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
DermAssist AI: AI-Powered Skin Disease Detection and Diagnostic Assistance System Using Deep Learning
|
| Subject Category |
Computer Engineering
|
| Abstract |
DermAssist AI is an advanced artificial intelligence-powered dermatological diagnostic assistance system that leverages deep learning and computer vision to analyse skin lesion images and provide preliminary diagnostic insights for a wide spectrum of common skin conditions. The system is built upon a Convolutional Neural Network (CNN) architecture enhanced with transfer learning from a pre-trained EfficientNetB3 model, trained on the HAM10000 dataset containing over ten thousand labelled dermatoscopic images spanning seven diagnostic categories: Melanocytic nevi, Melanoma, Benign keratosis-like lesions, Basal cell carcinoma, Actinic keratoses, Vascular lesions, and Dermatofibroma. The complete data science and software engineering lifecycle is implemented, encompassing systematic data preprocessing, augmentation, model training, performance evaluation using medical-grade metrics (AUC, sensitivity, specificity), and deployment as an interactive Flask web application. An explainability layer using Gradient-weighted Class Activation Mapping (Grad-CAM) highlights the specific image regions most influential in the diagnostic prediction. The system achieves a macro-averaged AUC of 0.889 across all seven classes, demonstrating strong generalisation capability. DermAssist AI represents a meaningful contribution to the democratisation of dermatological care through artificial intelligence.
|
| Keyword |
Skin Disease Detection, Deep Learning, EfficientNet, Transfer Learning, Grad-CAM, HAM10000, Dermatology AI, Flask Deployment, Medical Image Analysis, CNN
|
|
| Paper ID |
IJIFR/V13/E8/033
|
| Author |
Abburi Mahesh, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
M.Gowthami, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
IntentSense AI: Transformer-Based Customer Intent Detection System
|
| Subject Category |
Computer Engineering
|
| Abstract |
Natural Language Understanding (NLU) has become a cornerstone of modern artificial intelligence systems, enabling machines to interpret and respond to human language effectively. Among its critical components, intent detection plays a vital role in identifying the underlying purpose of user inputs in applications such as customer support automation, conversational agents, and virtual assistants. However, traditional approaches such as rule-based systems and recurrent neural networks often fail to capture contextual nuances and long-range dependencies in language. This paper introduces IntentSense AI, a Transformer-based customer intent detection system that leverages sequence-to-sequence learning using a bilingual English-Hindi parallel corpus. The proposed system utilizes multi-head self-attention mechanisms to efficiently model contextual relationships between words, overcoming the limitations of sequential processing models. The architecture is implemented using PyTorch and consists of embedding layers, a simplified Transformer encoder-decoder structure, and a linear projection layer. The model is trained for 50 epochs using the Adam optimizer and cross-entropy loss function, ensuring stable convergence. Furthermore, the system is deployed using Streamlit to enable real-time inference and user interaction. Experimental results demonstrate that the proposed approach achieves improved generalization and adaptability compared to traditional rule-based systems, making it suitable for practical deployment in customer service environments.
|
| Keyword |
Natural Language Understanding; Intent Detection; Transformer Architecture; Sequence-to-Sequence Learning; Bilingual Corpus; PyTorch; Streamlit; Customer Service Automation
|
|
| Paper ID |
IJIFR/V13/E8/032
|
| Author |
Gorrela Kaveri, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
M.Gowthami, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr.S.Usharani, Professor & Head, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
Real-Time Driver Fatigue Detection Using Eye Aspect Ratio and Mouth Aspect Ratio Analysis via Facial Landmark Localization
|
| Subject Category |
Computer Engineering
|
| Abstract |
Driver fatigue constitutes one of the most critical, yet frequently underestimated, contributors to fatal road traffic accidents worldwide. Epidemiological studies conducted by the World Health Organization and regional road safety authorities consistently attribute fifteen to twenty percent of motorway collisions to drowsiness-related impairment, with the actual incidence likely higher owing to systematic under-reporting. The principal challenge is that fatigue onset is gradual and subjective, rendering drivers incapable of reliably self-assessing their level of impairment, particularly during extended sessions of monotonous driving. Existing intervention strategies—encompassing hours-of-service legislation, electronic logging devices, and lane departure warning systems—operate at the policy or vehicle-dynamics level, and are intrinsically incapable of detecting the physiological precursors of drowsiness in real time.
This paper presents DriveSafe Vision, a non-intrusive, camera-based, real-time driver fatigue detection system implemented entirely in Python using open-source libraries. The proposed system continuously acquires video from a standard webcam and applies dlib's 68-point facial landmark shape predictor to localize anatomical reference points around the ocular and perioral regions in each frame. Two normalized geometric metrics are computed per frame: the Eye Aspect Ratio (EAR), formulated as the ratio of summed vertical inter-landmark distances to twice the horizontal inter-landmark distance across both eyes, and the Mouth Aspect Ratio (MAR), formulated analogously for the outer lip contour. Drowsiness is inferred when the EAR drops below a calibrated threshold of 0.25 for a sustained window of twenty or more consecutive frames, while yawning is detected when the MAR exceeds 0.60 for fifteen or more consecutive frames.
The system incorporates a heads-up display presenting session duration, cumulative blink and yawn counts, and animated EAR/MAR progress bars with threshold markers. Upon detection of a fatigue event, an unmistakable translucent red visual overlay and an optional audio alarm are simultaneously activated. Evaluation on a controlled volunteer cohort demonstrates a drowsiness detection precision of 91.3%, a recall of 88.7%, an F1-score of 89.98%, and a mean per-frame processing latency of 38.4 milliseconds on a mid-range consumer CPU, corresponding to an effective monitoring rate of approximately 26 frames per second. The system requires no proprietary hardware and is deployable on any platform supporting Python 3.8 or later, positioning it as a viable fatigue monitoring solution for commercial fleet operators, driving simulator research, and individual vehicle retrofit applications.
|
| Keyword |
Driver fatigue detection; Eye Aspect Ratio; Mouth Aspect Ratio; Facial landmark localization; Computer vision; Real-time monitoring
|
|
| Paper ID |
IJIFR/V13/E8/031
|
| Author |
Karamala Sai Pravallika, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
V. Vijayalakshmi, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
A MULTI -MODAL FINANCIAL PREDICTION SYSTEM COMBINING TRANSFORMER-BASED SENTIMENT ANALYSIS AND TECHNICAL INDICATORS
|
| Subject Category |
Computer Engineering
|
| Abstract |
Financial markets are driven by both quantitative fundamentals and qualitative information flows — news articles, analyst commentary, press releases, and regulatory announcements — that collectively shape market participant sentiment and govern short-term price dynamics. Translating unstructured financial text into structured predictive signals constitutes a central challenge in computational finance. This paper presents FinSent AI, an end-to-end, modular machine learning pipeline that ingests live financial news from RSS feeds, applies FinBERT — a BERT-based transformer language model pre-trained on financial text corpora — to derive continuous daily sentiment scores, and combines those scores with classical technical price features to train classification models predicting next-day stock price directional movement.
The system is demonstrated using Apple Inc. (AAPL) as the primary ticker, sourcing historical OHLCV data via the Yahoo Finance API and news headlines from Yahoo Finance and Reuters RSS feeds. Text preprocessing eliminates noise including URLs, punctuation, and stopwords, after which FinBERT classifies each article as positive, negative, or neutral with an associated confidence score. Daily aggregation of per-article signed sentiment scores — computed as the product of polarity label and model confidence — yields a continuous sentiment signal temporally aligned to the stock price time series.
Feature engineering yields a seven-dimensional input matrix comprising closing price, trading volume, daily return, five-day and ten-day moving averages, five-day rolling return volatility, and the daily sentiment score. Two ensemble classifiers — Random Forest and XGBoost — are trained on an 80/20 chronological train-test split to prevent data leakage. The superior model is deployed through an interactive four-tab Streamlit web dashboard delivering stock and sentiment visualization, prediction overlays, news browsing, and real-time next-day directional forecasts with confidence scores. The complete architecture is reproducible, extensible to any tradeable ticker, and readily deployable on cloud infrastructure.
|
| Keyword |
Financial Sentiment Analysis; FinBERT; Ensemble Learning; Stock Price Prediction; Natural Language Processing
|
|
| Paper ID |
IJIFR/V13/E8/030
|
| Author |
Akula Tharun, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr. S. Usharani, Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
MediFlow Optimizer: Hospital Resource & Patient Flow Intelligence System
|
| Subject Category |
Computer Engineering
|
| Abstract |
Hospital overcrowding, rooted in a systemic mismatch between highly dynamic patient demand and coarse-grained resource allocation practices, represents a critical patient safety and operational efficiency challenge in contemporary acute-care environments. Existing hospital management systems are predominantly retrospective in orientation, recording historical events without providing the predictive intelligence necessary for proactive resource deployment. MediFlow Optimizer addresses this fundamental gap through the design and implementation of a four-tier, machine learning-augmented hospital resource and patient flow intelligence platform. The proposed system integrates an Excel-based data acquisition layer, a Python-driven analytics engine employing a Random Forest Regressor (100 estimators) trained on temporal features—hour of day, day of week, and weekend indicator—extracted from historical admission records, a MySQL relational database persistence layer comprising three primary tables and three pre-aggregated analytical views, and a Microsoft Power BI interactive dashboard layer delivering operational intelligence to hospital administrators and clinical quality managers. The Random Forest model generates hourly patient inflow forecasts at seven-day granularity with a Mean Absolute Error (MAE) of approximately 0.85 patients per hour, a Root Mean Squared Error (RMSE) of 1.20, and a coefficient of determination (R²) of 0.78, outperforming baseline Linear Regression, standalone Decision Tree, and Gradient Boosting alternatives across all three evaluation metrics. Each forecasted hourly period is accompanied by peak-status classification (Normal, High, or Critical Peak) and actionable resource recommendations specifying appropriate clinical staffing ratios and bed deployment targets. The system further computes and tracks four daily key performance indicators—average patient wait time, total admission volume, high-severity case count, and bed utilization rate—presented through a custom dark-themed Power BI dashboard designed for operational deployment in clinical monitoring environments. The complete platform is implemented exclusively with open-source technologies, rendering it reproducible and accessible to healthcare organizations without investment in proprietary analytical infrastructure.
|
| Keyword |
Hospital Patient Flow Prediction; Random Forest Regressor; Healthcare Resource Allocation; Business Intelligence Dashboard; Time-Series Forecasting
|
|
| Paper ID |
IJIFR/V13/E8/029
|
| Author |
Bajanthri Reddy Kishore, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr. S. Usharani, Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
POWERPULSE ANALYTICS: A MACHINE LEARNING–DRIVEN FRAMEWORK FOR SMART ENERGY DEMAND FORECASTING AND LOAD OPTIMIZATION USING RANDOM FOREST REGRESSION
|
| Subject Category |
Computer Engineering
|
| Abstract |
Contemporary power systems are subject to increasingly volatile consumption patterns driven by urbanization, industrialization, and the proliferation of smart devices, exposing the limitations of conventional statistical forecasting methodologies. This paper presents PowerPulse Analytics, an end-to-end intelligent energy demand forecasting and load optimization framework that integrates ensemble machine learning with a structured data pipeline and interactive visualization. The system employs a Random Forest Regressor trained on a multi-dimensional dataset encompassing temporal attributes, regional consumer segmentation, ambient environmental variables (temperature and humidity), and holiday indicators to generate hourly energy consumption forecasts over rolling seven-day horizons. The architecture is organized into five functional layers: Data Acquisition, Feature Engineering and Preprocessing, Machine Learning Inference, Persistent Storage via a MySQL relational database, and Decision-Support Visualization through a Power BI dashboard. Feature engineering transforms raw timestamps into discriminative temporal signals including hour-of-day, day-of-week, and month, which are critical for capturing diurnal and seasonal consumption cycles. Experimental validation demonstrates that the Random Forest model achieves superior predictive accuracy compared to baseline statistical methods, with a Mean Absolute Error (MAE) below 4.2 kWh and an R² coefficient exceeding 0.91 across residential, commercial, and industrial consumer segments. The automated, modular pipeline architecture ensures reproducibility, scalability, and seamless integration with relational database infrastructure. The proposed framework provides utility operators with actionable decision-support intelligence to proactively mitigate demand spikes, reduce grid imbalances, and optimize resource allocation. Results demonstrate that machine learning-driven forecasting constitutes a substantively superior alternative to conventional heuristics, establishing a scalable blueprint for smart grid energy management systems.
|
| Keyword |
Energy Demand Forecasting; Random Forest Regressor; Smart Grid Optimization; Temporal Feature Engineering; Decision-Support Analytics
|
|
| Paper ID |
IJIFR/V13/E8/028
|
| Author |
Chakali Gireesh, Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
Dr. S. Usharani, Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
AN INTELLIGENT BANKING CUSTOMER CHURN PREDICTION AND LIFETIME VALUE ANALYTICS SYSTEM USING MACHINE LEARNING AND SURVIVAL MODELS
|
| Subject Category |
Computer Engineering
|
| Abstract |
The persistent challenge of banking customer churn imposes substantial revenue attrition on financial institutions operating within hyper-competitive digital environments. This paper presents Retain360, a comprehensive end-to-end analytical platform that addresses this challenge through the systematic integration of ensemble classification, survival analysis, and explainable artificial intelligence (XAI). The system is trained and evaluated on the IBM Banking Customer Churn Dataset, encompassing demographic, transactional, and service-usage attributes for 7,043 customer records. A Random Forest classifier, trained with class-balanced weighting and hyperparameter-optimised via Grid Search Cross-Validation, achieves an F1-Score of approximately 0.62 and a ROC-AUC of 0.85 on the held-out test partition — demonstrating discriminative capability substantially exceeding random baselines and competitive with state-of-the-art benchmarks. The survival analysis component employs the lifelines library to fit Kaplan-Meier survival curves and a Cox Proportional Hazards (Cox PH) model, enabling the derivation of individualised hazard functions and survival probability trajectories over customer tenure. These temporally-resolved risk profiles underpin a personalised Customer Lifetime Value (CLTV) estimator that translates survival-derived expected tenure into quantitative revenue projections. Model interpretability is achieved through a tri-layer explainability framework comprising Permutation Importance for global feature ranking, Partial Dependence Plots (PDPs) for marginal effect visualisation, and SHAP (SHapley Additive exPlanations) force plots for instance-level prediction attribution. The complete system is operationalised as a Flask-based web application delivering real-time churn probability scores, risk gauges, SHAP explanations, and survival visualisations through a form-driven interface accessible to non-technical banking professionals. Retain360 thus bridges the methodological gap between academic machine learning research and actionable, production-grade customer retention intelligence.
|
| Keyword |
Customer Churn Prediction; Random Forest; Cox Proportional Hazards; SHAP Explainability; Customer Lifetime Value
|
|
| Paper ID |
IJIFR/V13/E8/027
|
| Author |
A. Pramod Kumar, MCA Student, Department of Computer Applications, Viswam Engineering College, Andhra Pradesh, India
Dr. S. Usharani, Professor, Department of MCA, Viswam Engineering College, Madanapalli, Andhra Pradesh, India
|
| Paper Title |
AgroYield AI: An Interpretable Machine Learning Framework for Crop Yield Prediction in Precision Agriculture
|
| Subject Category |
Computer Engineering
|
| Abstract |
Accurate crop yield prediction is a critical component in ensuring sustainable agricultural development and efficient resource utilization. Traditional estimation methods, primarily based on manual surveys and empirical observations, are often limited by scalability, time constraints, and their inability to model complex interactions among agronomic variables. This paper presents AgroYield AI, an interpretable machine learning framework for crop yield prediction using multi-dimensional agricultural data. The proposed system leverages a Random Forest regression model trained on diverse features including crop type, seasonal variations, geographical location, rainfall, fertilizer usage, and pesticide application. A robust preprocessing pipeline incorporating categorical encoding and feature normalization is employed to enhance predictive performance. The system integrates predictive modeling with an interactive visualization interface developed using Streamlit and D3.js, enabling real-time data exploration and decision support. Experimental evaluation demonstrates that the proposed model achieves high predictive accuracy and strong generalization capability compared to conventional regression approaches. Furthermore, feature importance analysis provides transparency and interpretability, allowing stakeholders to understand the influence of key agricultural parameters on yield outcomes. The proposed framework offers a scalable and practical solution for data-driven agricultural decision-making, supporting farmers, researchers, and policymakers.
|
| Keyword |
Crop Yield Prediction; Machine Learning; Random Forest; Precision Agriculture; Data Analytics
|
|
| Paper ID |
IJIFR/V13/E8/026
|
| Author |
Bonasi Anusha, MCA Student, Department of Computer Applications, Viswam Engineering College, Andhra Pradesh, India
Dr. S. Usharani, Professor, Department of MCA, Viswam Engineering College, Madanapalli, Andhra Pradesh, India
|
| Paper Title |
Smart AI-Based Framework for Automated Legal Clause Risk Assessment in Contracts
|
| Subject Category |
Computer Engineering
|
| Abstract |
In the modern legal and corporate environment, contract review remains a critical yet resource-intensive task, often requiring significant time, expertise, and manual effort. Traditional approaches to reviewing legal documents are prone to human error, inconsistencies, and inefficiencies, especially when dealing with large volumes of contracts. This paper presents ContractRisk Analyzer AI, an intelligent system designed to automate the identification and classification of legal risk within contract documents using Natural Language Processing (NLP) techniques. The proposed system processes legal contracts in both PDF and plain-text formats, extracting and segmenting the content into individual clauses using advanced sentence tokenization methods. Each clause is evaluated against a structured legal risk knowledge base categorized into High, Medium, and Low risk levels. The classification process is driven by a rule-based NLP engine that ensures transparency by identifying the exact phrases responsible for risk detection. A weighted scoring mechanism aggregates clause-level risks to generate an overall document risk percentage, providing a clear and interpretable assessment. Furthermore, the system integrates an intelligent recommendation module that offers actionable insights based on detected risks, assisting users in making informed legal decisions. Implemented as a web-based application using the Flask framework, the system provides an intuitive user interface with interactive visualizations, including a dynamic risk gauge and clause-level analysis tables. The proposed solution demonstrates the effectiveness of explainable AI in the legal domain, offering a scalable, accessible, and efficient alternative to traditional contract review processes. It significantly reduces review time while maintaining analytical accuracy, making it valuable for legal professionals, organizations, and academic users.
|
| Keyword |
Contract Analysis; Natural Language Processing; Legal Risk Assessment; Clause Classification; Explainable AI
|
|
| Paper ID |
IJIFR/V13/E8/025
|
| Author |
Kodavati Mounika, MCA Student, Department of Computer Applications, Viswam Engineering College, Andhra Pradesh, India
B. Shreesha, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Andhra Pradesh, India
Dr. Usha Rani, Professor, Department of Computer Applications, Viswam Engineering College,Madanapalli,Andhra Pradesh
|
| Paper Title |
Creditshield AI: An Explainable Machine Learning Framework for Loan Default Risk Prediction
|
| Subject Category |
Computer Engineering
|
| Abstract |
Credit risk assessment plays a vital role in ensuring the financial stability and sustainability of lending institutions. Traditional credit scoring methods, primarily based on statistical models such as logistic regression, often fail to capture complex, non-linear relationships inherent in borrower data. This limitation results in reduced predictive performance, especially in dynamic financial environments. To address these challenges, this paper presents CreditShield AI, an explainable machine learning-based loan default risk prediction system designed to enhance accuracy, transparency, and reliability in credit decision-making.The proposed system leverages the Give Me Some Credit dataset, comprising over 150,000 borrower records with multiple financial and behavioral attributes. A structured data pipeline is developed, including data preprocessing, missing value imputation, outlier detection, feature scaling, and exploratory data analysis. Robust statistical techniques such as median imputation and percentile-based winsorization are applied to ensure data quality and consistency. Furthermore, the system adopts Robust Scaler normalization to mitigate the impact of extreme values.A key contribution of this work is the emphasis on explainable AI, ensuring that model predictions can be interpreted in compliance with financial regulatory standards. The system is designed to integrate advanced machine learning models and interpretability techniques such as SHAP and LIME in future stages. The proposed framework not only improves prediction capability but also promotes fairness, transparency, and responsible AI practices in financial risk management.
|
| Keyword |
Credit Risk Prediction; Machine Learning; Loan Default; Explainable AI; Financial Analytics
|
|
| Paper ID |
IJIFR/V13/E8/024
|
| Author |
K. Bhavani Sankar, MCA Student, Department of Computer Applications, Viswam Engineering College, Andhra Pradesh, India
V.Vijayalakshmi, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Andhra Pradesh, India
Dr. Usha Rani, Professor, Department of Computer Applications, Viswam Engineering College,Madanapalli,Andhra Pradesh
|
| Paper Title |
DisasterVision AI: A Satellite-Based Deep Learning Framework for Automated Disaster Damage Assessment
|
| Subject Category |
Computer Engineering
|
| Abstract |
Natural disasters such as earthquakes, floods, hurricanes, and wildfires cause extensive damage to infrastructure and human life, making rapid and accurate damage assessment a critical requirement for effective disaster response. Traditional ground-based assessment techniques are time-consuming, risky, and limited in spatial coverage, which delays emergency decision-making processes. To address these limitations, this paper presents DisasterVision AI, an automated satellite imagery analysis system that leverages deep learning for large-scale building damage assessment. The proposed system utilizes a modified Single Shot MultiBox Detector (SSD) with a VGG-16 backbone, enhanced to process six-channel input by combining pre-disaster and post-disaster satellite images. This dual-input architecture enables the model to learn visual differences between temporal image pairs, improving damage detection accuracy. The model is trained using the xView2 dataset, which provides annotated satellite imagery with four damage categories: no-damage, minor-damage, major-damage, and destroyed. The system incorporates advanced training techniques including data augmentation using Albumentations, OneCycle learning rate scheduling, and AdamW optimization for efficient convergence. Performance evaluation is conducted using Mean Average Precision (mAP) metrics across multiple IoU thresholds. Additionally, Non-Maximum Suppression (NMS) is applied for refining detection outputs. Experimental results demonstrate that DisasterVision AI provides fast, scalable, and reliable damage assessment, making it a valuable tool for disaster management authorities and emergency response teams.
|
| Keyword |
Disaster Damage Assessment; Deep Learning; Satellite Imagery; Object Detection; SSD; xView2 Dataset
|
|
| Paper ID |
IJIFR/V13/E8/023
|
| Author |
Sirisani Lavanya, M.C.A. Student, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
M. Gowthami, Assistant Professor, Department of Computer Applications, Viswam Engineering College, Madanapalle, Andhra Pradesh, India
|
| Paper Title |
TrafficOpt RL: Adaptive Traffic Signal Optimization Using Deep Reinforcement Learning
|
| Subject Category |
Computer Application & Engineering
|
| Abstract |
Urban traffic congestion is a critical infrastructure challenge facing modern cities as vehicle populations expand and urban density increases. Conventional fixed-timing traffic signal systems are incapable of adapting to the stochastic and dynamic nature of real-world traffic flows, resulting in wasted green-light time, queue buildup, increased vehicle emissions, and emergency response delays. This paper presents TrafficOpt RL, an end-to-end adaptive traffic signal optimization system that applies the Deep Q-Network (DQN) algorithm to learn intelligent signaling policies at urban intersections through iterative simulation experience. The system is built on a custom Gymnasium-compatible simulation environment modeling a four-way intersection with stochastic Poisson vehicle arrivals. The DQN agent, implemented via the Stable-Baselines3 framework, utilizes experience replay, target network stabilization, and epsilon-greedy exploration to converge on policies minimizing aggregate vehicle waiting times and maximizing intersection throughput. All training metrics and simulation data are persistently stored in a MySQL relational database through automated callback logging, enabling systematic performance analysis. Evaluation via direct comparison against a fixed-timing baseline demonstrates measurable superiority of the reinforcement learning approach across three performance dimensions: average vehicle waiting time, total throughput, and composite efficiency score. Three analytical visualizations are generated to communicate system performance. TrafficOpt RL constitutes a practical proof-of-concept for deep reinforcement learning integration into intelligent transportation systems and smart city infrastructure.
|
| Keyword |
Deep Reinforcement Learning; Traffic Signal Optimization; Deep Q-Network; Intelligent Transportation Systems; Adaptive Control
|
|
| Paper ID |
IJIFR/V13/E8/022
|
| Author |
A. SARALA, Research Scholar, Alagappa University College of Education, Alagappa University, Karaikudi
Dr. J.E. MERLIN SASIKALA, Associate Professor, Alagappa University College of Education, Alagappa University, Karaikudi
|
| Paper Title |
A STUDY ON THE EFFECTIVENESS OF BLENDED LEARNING ON THE ACADEMIC ACHIEVEMENT OF HIGHER SECONDARY STUDENTS
|
| Subject Category |
EDUCATION
|
| Abstract |
The present study investigates the impact of blended learning on the academic performance and learning attitudes of higher secondary students. A total of 100 students from two higher secondary schools were selected through purposive sampling. The results reveal a noticeable difference in achievement scores between students exposed to blended learning and those taught through conventional methods, favouring the blended approach. Blended learning refers to the integration of classroom teaching with digital learning tools such as online discussions, multimedia content, and interactive activities. This approach supports flexible learning, enhances engagement, and allows students to learn at their own pace. The findings indicate that students who participated in blended learning demonstrated better academic outcomes compared to those in traditional settings. Therefore, incorporating blended strategies at the higher secondary level can significantly improve learning effectiveness and student engagement.
|
| Keyword |
Blended Learning, Academic Achievement, Higher Secondary Students, Learning Effectiveness
|
|
| Paper ID |
IJIFR/V13/E8/020
|
| Author |
K. Baswaraj, Assistant Professor
S. M. Joshi, Professor
|
| Paper Title |
Proactive Ransomware Detection and Mitigation by Demonstrating the Advantages of Integrating AI and ML in Cyber security
|
| Subject Category |
Computer Engineering
|
| Abstract |
The core component of the suggested AI-driven ransomware detection system is anomaly detection. One may effectively identify ransomware operations by utilizing anomaly detection algorithms, which can define baseline behaviors and detect deviations from these standards [9]. Artificial intelligence systems can separate benign and dangerous actions using clustering, statistical analysis, and outlier detection methods. Consequently, it responds quickly and reduces the occurrence of false positives. The AI framework's categorizing algorithms help find and classify abnormalities precisely. AI may facilitate the identification of ransomware attacks by automating the analysis of vast quantities of data. Machine learning models acquire knowledge from assault data that has already transpired to detect potential hazards in real time. The scalability, flexibility, and detection accuracy of ransomware strategies that are constantly evolving are enhanced by the implementation of AI. The system can discriminate between benign and suspicious behavior, suggesting ransomware using supervised learning methods such as support vector machines and random forests. By using classifiers trained on labeled datasets, including different strains of ransomware and typical behaviors, the artificial intelligence system could learn to recognize new risks and improve its skills. Artificial intelligence and machine learning-driven ransomware detection solutions improve detection accuracy and efficacy and enable proactive security measures. Artificial intelligence (AI) driven systems can adapt to fresh data and changing threat scenarios in real-time, averting ransomware events. The study establishes a firm foundation for proactive ransomware detection and mitigation by demonstrating the advantages of integrating AI and ML in cybersecurity.
|
| Keyword |
Ransomware Attacks; Machine Learning; Anomaly Detection; Threat Migration
|
|
| Paper ID |
IJIFR/V13/E8/019
|
| Author |
Dr Ravita Kanan, Professor, pgimer
Dr Raghvendra S, Professor, pgimer
|
| Paper Title |
Deep visual detection system for oral squamous cell carcinoma
|
| Subject Category |
Oncology
|
| Abstract |
Oral cancer remains one of the most prevalent forms of cancer globally, ranking sixth among the most common types. Specifically, oral squamous cell carcinoma (OSCC) constitutes nearly 90% of the aggressive cases, posing significant health risks to affected individuals1. The World Health Organization (WHO) highlights that approximately 657,000 new cases of oral cancer are diagnosed each year globally, leading to more than 330,000 fatalities annually. This alarming statistic underscores the severe impact of OSCC, especially prevalent in developing countries across South and Southeast Asia, where incidence rates are nearly double the global average. India, in particular, accounts for one-third of the global OSCC cases, emphasizing a substantial healthcare burden in the region. Data augmentation techniques were employed to mitigate class imbalance and enhance model generalization, while advanced image pre-processing methods and training strategies such as EarlyStopping and Reduce LROn Plateau were applied to ensure stable convergence. Results Among the models tested, EfficientNetB3 consistently delivered superior performance across both datasets. On the binary classification task, it achieved a test accuracy of 97.05%, with precision, recall, and F1-score all at 97.05%, specificity of 97.17%, and sensitivity of 96.92%. On the multi-class NDB-UFES dataset, it again outperformed the other models, attaining a 97.16% accuracy, matching precision, recall, and F1-score, and specificity of 98.58%. In contrast, DenseNet121 and ResNet50 showed substantially lower accuracy scores in both experiments.
|
| Keyword |
Kaggle OSCC dataset, NDB-UFES Oral Cancer Dataset, Deep Visual Detection System Training Strategy, OSCC Image Classification.
|
|
| Paper ID |
IJIFR/V13/E8/018
|
| Author |
Romya Sharon Immadi, Mount Carmel College Autonomous, Bangalore
Hamsa N, Mount Carmel College Autonomous Bangalore
|
| Paper Title |
Big Five Personality Traits and Loneliness among Siblings of Individuals with Intellectual Disability
|
| Subject Category |
Psychology
|
| Abstract |
The complex role of a sibling in long-term caregiving and their experienced psychological distress is often overlooked by stakeholders, interventions, rehabilitative efforts and support systems for individuals with Intellectual Disabilities.The study explores the Big Five Personality Traits profile and loneliness using the Revised NEO Five Factor Inventory, and UCLA Loneliness Scale version 3, including the correlation between them. 30 adult siblings of the age group 18-51 participated in the study. The data obtained were subjected to the test of normality, descriptive statistics, t-test, and correlation analysis. High levels of Neuroticism, Low Extraversion, Average Openness to Experience, Low Agreeableness, Low Conscientiousness, and Moderately High levels of Loneliness among siblings of people with Intellectual Disability. Results suggest there is a significant relationship between Neuroticism, Extraversion and Loneliness individually. The findings provide significant insights into the traits and mental health state, having implications for developing psychological and community support systems.
|
| Keyword |
Keywords: Personality, Neuroticism, Extraversion, Loneliness, Intellectual Disability, Sibling
|
|
| Paper ID |
IJIFR/V13/E8/017
|
| Author |
H. Lalrinhlui, Faculty of Disability Management and Special Education Ramakrishna Mission Vivekananda Educational and Research Institute Coimbatore, Tamil Nadu
Dr. S. Parween, Faculty of Disability Management and Special Education Ramakrishna Mission Vivekananda Educational and Research Institute Coimbatore, Tamil Nadu
|
| Paper Title |
Impact of Digital Literacy Training Programme on Pre-Employability Skills among Students with Visual Impairment
|
| Subject Category |
Special Education
|
| Abstract |
Students with visual impairment face persistent structural and technological barriers that hinder the development of employability-related skills in higher education. Despite inclusive policy frameworks such as the Rights of Persons with Disabilities (RPwD) Act, gaps remain in equipping these students with essential digital competencies for workforce participation. This study examined the effectiveness of a structured Digital Literacy Training Programme in enhancing pre-employability skills among students with visual impairment. An experimental pre-test-post-test design was employed, involving students with visual impairment divided into experimental and control groups. Data were analyzed using appropriate non-parametric statistical techniques. The findings revealed a significant improvement in pre-employability skills among students who received the training, whereas no notable change was observed in the control group. The results confirm the positive impact of structured digital literacy interventions on skill development. The findings emphasize the need for strong institutional commitment to integrating accessible digital training, assistive technologies, and inclusive pedagogical practices.
|
| Keyword |
Digital literacy, pre-employability skills, higher education, workforce readiness, students with visual impairment
|
|
| Paper ID |
IJIFR/V13/E8/016
|
| Author |
T. Akila, Vellalar college for women (Autonomous), Erode
V.Tamilselvi, Vellalar college for women (Autonomous), Erode
|
| Paper Title |
VERTEX AND EDGE BASED SPLIT, NON-SPLIT AND INVERSE DOMINATION NUMBERS OF HELM, JELLYFISH, SUNFLOWER AND FLOWER GRAPHS
|
| Subject Category |
Graph Theory -Mathematics
|
| Abstract |
This project investigates of various domination number on four specific graph families such as Helm graphs, Jellyfish graphs, Sunflower graphs and Flower graphs. The analysis covers both vertex and edge based domination, with particular emphasis on variants such as split domination, non-split domination, inverse split domination and inverse non-split domination. For each graph family, dominating sets are constructed and examined to determine their corresponding domination numbers. These findings confirm the presence of domination measures within each graph structure and establish their numerical values. This research properties contributes to the broader field in domination theory by extending the concepts to more complex and specialized graph families, providing useful insights into their structural properties.
|
| Keyword |
Domination number, split and non-split domination, inverse domination, special graph classes
|
|
| Paper ID |
IJIFR/V13/E8/015
|
| Author |
T.Akila, Vellalar college for women (Autonomous), Erode
V.Tamilselvi, Vellalar college for women (Autonomous), Erode
|
| Paper Title |
VERTEX AND EDGE BASED SPLIT, NON-SPLIT AND INVERSE DOMINATION NUMBERS OF HELM, JELLYFISH, SUNFLOWER AND FLOWER GRAPHS
|
| Subject Category |
Graph Theory
|
| Abstract |
This project investigates of various domination number on four specific graph families such as Helm graphs, Jellyfish graphs, Sunflower graphs and Flower graphs. The analysis covers both vertex and edge based domination, with particular emphasis on variants such as split domination, non-split domination, inverse split domination and inverse non-split domination. For each graph family, dominating sets are constructed and examined to determine their corresponding domination numbers. These findings confirm the presence of domination measures within each graph structure and establish their numerical values. This research properties contributes to the broader field in domination theory by extending the concepts to more complex and specialized graph families, providing useful insights into their structural properties.
|
| Keyword |
Domination number, split and non-split domination, inverse domination, special graph classes
|
|
| Paper ID |
IJIFR/V13/E8/014
|
| Author |
K. RAMESH KALE
S. M WAGHMARE
|
| Paper Title |
Power Management in Ad Hoc Networks by route redundancy in dense network by turning off devices
|
| Subject Category |
Computer Networks Engineering
|
| Abstract |
Ad hoc networking allows portable mobile devices to establish communication path without having any central infrastructure. Since there is no central infrastructure and the mobile devices are moving randomly, gives rise to various kinds of problems, such as routing and security. In this paper the problem of routing is considered. Wireless network devices, especially in ad hoc networks, are typically battery powered. The growing need for energy efficiency in wireless networks, in general, and in mobile Adhoc networks (MANETs), in particular, calls for power enhancement features. Battery power is an important resource in ad hoc networks. It has been observed that in ad hoc networks, energy consumption does not reflect the communication activities in the network. With the proliferation of portable computing platforms and small wireless devices, ad hoc wireless networks have received more and more attention as a means for providing data communications among devices regardless of their physical locations. Wireless communication has the advantage of allowing untethered communication, which implies reliance on portable power sources such as batteries. However, due to the slow advancement in battery technology, battery power continues to be a constrained resource and so power management in wireless networks remains to be an important issue. It has been observed that in ad hoc networks, energy consumption does not always reflect active communication in the network [1]. Experimental results reveal that the energy consumption of wireless devices in an idle state is only slightly smaller than that in a transmitting or receiving state. Therefore, it is in general desirable to turn the radio off when it is not in use. Motivated by these observations, several energy conservation protocols [2], [3] have been proposed to take advantage of route redundancy in dense ad hoc networks by turning off devices that are not required for global network connectivity. However, in these protocols, the decision about which set of nodes to leave on is only based on geographical/topological information thus is oblivious to the actual traffic load in the network.
|
| Keyword |
Ad Hoc Wireless Networks, Energy Conservation Protocols, Battery Power Observation, Power Enhancement Features
|
|
| Paper ID |
IJIFR/V13/E8/013
|
| Author |
Karthika Singaravelu, MASS College of Arts and Science, Kumbakonam
|
| Paper Title |
Effects of Artificial Intelligence on Academics in Higher Education in India
|
| Subject Category |
Education
|
| Abstract |
India's higher education system has undergone radical change in recent years, propelled in large part by the introduction of AI into the classroom. Artificial intelligence (AI), with its ability to crunch massive amounts of data, make predictions, and automate a wide range of tasks, has ushered in a new era in the classroom . Since it affects so many facets of higher education, from teaching methods to administrative duties, it is an essential topic of research for anybody interested in this sector. This investigation focuses on the implications of AI implementation for Indian college students. This covers a wide range of issues, such as how AI is used in schools, how it affects students' educational and professional paths, and what its future prospects may be. Personalized learning, automated grading systems, intelligent tutoring, and even administrative responsibilities like admissions and resource allocation are all included in the scope of this topic. The impact of artificial intelligence (AI) technology' increasing permeation into India's educational infrastructure is becoming an increasingly pressing topic of study. This research aims to examine the wide range of AI applications and evaluate the potential and threats it poses to today's students in the classroom. The significance of including AI-related courses and chances for skill development into the curriculum is brought into focus by the finding of the study that students have a positive outlook on their ability to compete in labour markets dominated by AI. Because both technology and education are always evolving, it is imperative that AI integration be subjected to continuous monitoring and feedback collection, as well as iterative improvements. Schools that take an adaptable approach such as this one are better able to satisfy the shifting demands of their pupils as well as the ever-evolving nature of technology.
|
| Keyword |
Artificial Intelligence, Quality Education, Indian Education System, Adaptive Learning Technology
|
|
| Paper ID |
IJIFR/V13/E8/012
|
| Author |
S K Sahil, Assistant Professor of Law, School of Legal Studies, Seacom Skills University, Birbhum, West Bengal, India.
|
| Paper Title |
DRAWBACKS AND ADVERSE LEGAL IMPLICATIONS OF SECTION 69 OF THE BHARATIYA NYAYA SANHITA 2023: A CRITICAL ANALYSIS
|
| Subject Category |
Law
|
| Abstract |
Section 69 of the Bharatiya Nyaya Sanhita, 2023 introduces a new offence addressing sexual intercourse obtained through deceitful means, particularly false promises of marriage and other inducements. While the provision aims to strengthen the legal framework for protecting women from sexual exploitation, it raises significant concerns regarding its interpretation and practical application. The section suffers from vague and ambiguous language, lack of clarity in defining “consent” and “deceitful means,” and the absence of comprehensive safeguards against misuse. It also reflects a genderbiased approach by recognizing only women as victims, thereby excluding men and LGBTQ+ individuals from its protection.This study adopts a doctrinal methodology, relying on primary sources such as statutory provisions and judicial decisions along with secondary sources including books, research articles, journals and credible electronic resources. Analytical, descriptive and exploratory approaches are used to critically examine the scope, limitationsand implications of the provision.The findings reveal that Section 69 of BNS, despite its progressive intent, creates legal uncertainty, difficulty in proving intention and potential for misuse, which may lead to inconsistent judicial outcomes. The study concludes that there is an urgent need for clearer judicial interpretation, precise legislative drafting, and balanced safeguards to ensure that the provision achieves its intended objective without compromising fairness and justice.
|
| Keyword |
Free Consent, Deceitful Means, False Promise of Marriage, Misuse of Law, Gender Biasness, Misuse of Law
|
|
| Paper ID |
IJIFR/V13/E8/011
|
| Author |
Dr Pravin Choudhary, Oriental College Of Management,Bhopal
Dr Richa Agnihotri, Oriental College of Management, Bhopal
Dr Shikha Bhargava, Oriental College of Management, Bhopal
|
| Paper Title |
A Study on Merger & Acquisitions and its impact on organizational change with special reference to Bank of Baroda
|
| Subject Category |
Finance
|
| Abstract |
This research examines the organizational transformation that followed the merger of Bank of Baroda with Vijaya Bank and Dena Bank on April 1, 2019, an integration that resulted in the formation of India’s third-largest public sector bank. The main objective is to analyze the various dimensions of organizational change by Merger & Acquisition, specifically focusing on the challenges and outcomes related to operation, human resource management, cultural alignment, and financial performance. Our study is based on primary and secondary data both. Secondary data which is taken from annual published report of selected bank web site and primary data which about the pre and post impact on organizational change which is collected through questionnaire distributed amongst the employees of selected bank. The study evaluates several key parameters, including Gross NPA, Net NPA, Net Interest Margin, Return on Assets, and Return on Equity. In addition, it examines the work culture and work environment of employees in the selected bank. We use the statistical tool for analysis is Mean, Standard deviation and also t-test and chi-square test used for hypothesis testing. We conclude that the impact of merger and acquisition on organization are positive.
|
| Keyword |
Organizational Change, Gross NPA, Net NPA, ROA, ROE and NIM
|
|
| Paper ID |
IJIFR/V13/E8/010
|
| Author |
Dr. Vaishali Kiran Ghadyalji, K. J. Somaiya Institute of Technology, Sion, Mumbai, M.S.-400022.
|
| Paper Title |
An Exploration of Human Malevolence: Juxtaposing Manthara and Shakuni
|
| Subject Category |
Humanities (English)
|
| Abstract |
Human nature is fundamentally malevolent as it is very natural for human beings to be selfish and narcissists. Humans are born with greed, envy and jealousy; and these innate feelings make them indulge into different kinds of immoral activities. On the contrary, their benevolence originates from the cognizant and constant imbibing of moral values. Many philosophers and scholars right from Aristotle to Thomas Hobbs, Charls Darwin, Sigmund Freud, Peter Muris, Herald Merckelbach, Henry Otgaar, David Coady, Lee Besham and a plethora have tried to explain and restate this fact. This paper is an attempt to explore the inexorable and unfortunate element of malevolence in human beings by juxtaposing the supposed villainous characters- Manthara from Ramayana and Shakuni from Mahabharata- with an analysis of their influential endevours to inspire their target audience to act according to their philosophy. Manthara, hunchbacked and old, was the confidante attendant and favourite maid of Queen Kaikeyi in the Indian epic Ramayana who instigated her to convince King Dasharatha to coronate Bharata in the place of Lord Rama and ask for fourteen years of exile to Lord Rama. Shakuni was the king of Gandhar and brother of Gandhari, the queen of Dhritrashtra and Hastinapur and the mother of hundred Kauravas in another Indian epic Mahabharata. Shakuni is delineated as one of the exceptionally intelligent characters in the epic but a very scheming one. This article endeavours to compare these two characters possessing malevolent traits, trying to establish that malevolence is not grounded in particular gender, position, era or such related aspects; and the malevolence of both of these characters does enclose certain grey shades.
|
| Keyword |
human malevolence, scheming, disability, manipulation, malicious counsel
|
|
| Paper ID |
IJIFR/V13/E8/009
|
| Author |
Dr. Arunidhita
Dr. Amita S Vendi
|
| Paper Title |
Influence of Sustainable Development Goals (SDGs) on Cause Related Marketing Strategy and Consumer Purchase
|
| Subject Category |
Commerce
|
| Abstract |
There are two emerging and important challenges faced by the organizations currently. One is to work towards fulfillment of the Sustainable Development Goals (SDGs) to ensure keeping in line with the global challenges of becoming sustainable. The second is to portray themselves as highly ethical and cause driven organization to appeal to customer’s and society’s ethical needs and expectations along with making profits by implementing Cause Related Marketing (CRM) strategies. This concept paper based on secondary research explores the notion of applying Cause related marketing strategies to achieve the Sustainable Development Goals by an organization.The research paper examines the influence of the Sustainable Development Goals (SDGs) on formulating
the Cause Related Marketing strategy adopted by organizations and its effect on the consumer’s intention to
purchase the products. The research paper identifies Cause Related Marketing constructs which can be associated
with sustainability and Triple Bottom Line. ‘People’ relates to the social aspect of CRM, ‘Planet’ relates to the
Environmental responsibility of the organization and ‘Profit’ relates to the economic aspect of the organization.
The paper also examines the effect of such an SDG-oriented Cause Related Marketing strategy on the consumer’s
purchase intention.
|
| Keyword |
Business, Cause related marketing, NGOs, Sustainable development goals.
|
|
| Paper ID |
IJIFR/V13/E8/008
|
| Author |
Dr. Suchitra, Raghvan
Dr. Charu, Verma
|
| Paper Title |
Cross-Cultural Digital Marketing Strategies in the Age of Globalization:Adapting Global Marketing Strategies for Local Success
|
| Subject Category |
Commerce
|
| Abstract |
When companies engage in business activities abroad, they face the options of standardization, adaptation, or both for their marketing strategies. Standardization means using the same marketing strategy across the globe. This strategy can provide cost reduction advantages and establish a consistent global brand and operational efficiencies. When customers in different countries perceive similar needs, businesses can present a uniform, meaningful brand image. On the other hand, with adaptation, companies modify the product or marketing strategies to accommodate the lifestyle of each community.The paper recommends companies adopt global and local strategies to enhance their effectiveness in various markets. Marketers are advised to conduct cultural research, collaborate with local partners, and implement flexible strategies to foster ongoing loyalty in other countries.
|
| Keyword |
Cross-cultural marketing; Consumer behavior; Global strategy; Cultural adaptation; Glocalization; International branding
|
|
| Paper ID |
IJIFR/V13/E8/007
|
| Author |
Ms.Liza Alex, Baselius College,Autonomous Kottayam
Prof.(Dr.)Mini Joseph, Kuriakose Gregorious College,Pampady
CA Reshma Rachel Kuruvilla, BCM College,Kottayam
|
| Paper Title |
Sustainable Performance as a Mediator between Green Banking Practices and Competitive Advantage—Evidence from Indian Private Banks
|
| Subject Category |
HRM
|
| Abstract |
Purpose
The objective of this article is to look into how competitive advantage and green banking practices associate with one another in the Indian banking industry. It considerably looks over how sustainable performance functions as a mediator in this relationship. This study (Siddik et al., 2024) aims to make clear how ecologically responsible actions result in strategic advantages for banks (Gunawan et al., 2022) by investigating the degree to which sustainable performance explains the impact of green banking practices on competitive advantage.
Design/methodology/approach
With the aim of examining the connection between eco-friendly banking practices
and competitive advantage in the Indian banks, this quantitative study uses structural equation modeling with SmartPLS 4.0, with a focus on sustainable performance as a mediator. Convenient sampling will be used to gather the primary dataset from Indian private banks. The causal relationship between GBPs and CA will be assessed using SEM, and the mediating function of SP in this relationship will also be ascertained.
Findings
According to the analysis, the banking industry's competitive advantage is increased by putting green banking practices into practice. Additionally, the relationship between green practices and competitive advantage is strengthened by sustainable performance, indicating that green initiatives are more purposively effective when banks demonstrate strong sustainable performance. According to the partial mediation finding, banks gain from green banking practices in two ways: they directly increase their competitive advantage and improve sustainable performance (Siddik et al., 2024).
Originality/value
By learning the circumstances in which green banking practices improve a company's competitiveness, this study contributes to the expanding corpus of research on sustainable banking. It spotlights the significance of sustainable performance as a strategic tool for enhancing the advantages of green banking as well as a desired result (Siddik et al., 2024).
|
| Keyword |
Green banking practices, sustainable performance, competitive advantage, mediation analysis, Indian banking sector.
|
|
| Paper ID |
IJIFR/V13/E8/006
|
| Author |
G.R. Yarasi, CMR College of Engineering & Technology, Hyderabad/Bhagyanagar, Telangana, India
V. Lakshmi Narasimhan, Elizabeth City State University, NC 27909, USA
D. Rammohan, CMR College of Engineering & Technology
|
| Paper Title |
AI-Driven Algorithms Over NDB-UFES Oral Cancer Histopathological Dataset
|
| Subject Category |
Computer Engineering & Medical AI
|
| Abstract |
Oral cancer remains a significant global health challenge, particularly in regions where limited access to specialized pathology services delays diagnosis and treatment. Histopathological analysis is the clinical gold standard for detecting oral squamous cell carcinoma (OSCC), yet manual examination is time-consuming and susceptible to inter-observer variability. This study investigates the application of artificial intelligence (AI) techniques for analyzing histopathological images from the NDB-UFES Oral Cancer dataset, which contains 237 hematoxylin–eosin stained biopsy images representing OSCC and leukoplakia cases with varying degrees of dysplasia. Three AI-driven approaches were evaluated: a baseline convolutional neural network (CNN) model for image classification, a generative adversarial network (GAN) framework for synthetic data augmentation, and a Vision Transformer (ViT) model for advanced image representation learning. The objective was to assess the feasibility and effectiveness of these architectures under realistic small-dataset conditions commonly encountered in medical imaging research. Experimental results demonstrate that while deep learning models such as CNNs and Vision Transformers have shown high performance in large-scale medical imaging tasks, their accuracy is substantially reduced when trained on limited datasets. In this study, the CNN model achieved an accuracy of 56.25%, while the Vision Transformer achieved 54.43%. These findings highlight the impact of dataset size on model generalization and emphasize the importance of robust data augmentation and transfer learning strategies. The study provides insights into the applicability of modern AI techniques for oral cancer histopathological analysis and underscores the need for larger annotated datasets and multimodal learning approaches to support reliable AI-assisted diagnostic systems in oral oncology.
|
| Keyword |
Artificial Intelligence,Oral Cancer Detection, Histopathological Image Analysis, Convolutional Neural Networks, Vision Transformer, Generative Adversarial Networks, NDB-UFES Dataset, Deep Learning,Medical Image Classification, Oral Squamous Cell Carcinoma
|
|
| Paper ID |
IJIFR/V13/E8/005
|
| Author |
G.Yarasi, CMR College of Engineering & Technology, Hyderabad/Bhagyanagar, Telangana, India
V. Lakshmi Narasimhan, Elizabeth City State University, NC 27909, USA
D. Rammohan, D. Rammohan
|
| Paper Title |
Chilli Image Classification Using ResNet50, VGG16 & InceptionV3 Algorithms – A Comparative Analytics
|
| Subject Category |
Computer Engineering
|
| Abstract |
This study presents a comparative evaluation of four deep learning architectures—CNN, ResNet50, VGG16, and InceptionV3—for chilli growth stage recognition and leaf disease classification. The models were assessed to determine their effectiveness in accurately distinguishing developmental stages and identifying disease conditions from image data. Experimental results demonstrate that transfer learning substantially improves classification performance across all architectures. Among the evaluated models, ResNet50 consistently achieved superior accuracy and overall performance in both growth stage and disease classification tasks. These findings highlight the effectiveness of deep transfer learning approaches in agricultural image analysis. Future research will focus on validating model robustness under real-field environmental conditions, designing lightweight architectures for mobile and edge deployment, and expanding the framework to support multi-crop classification systems for broader agricultural applications.
|
| Keyword |
Chilli Image Classification, ResNet50, VGG16 & InceptionV3 Algorithms, Comparative Analytics
|
|
| Paper ID |
IJIFR/V13/E8/004
|
| Author |
M. Patel, CMR College of Engineering & Technology, Hyderabad/Bhagyanagar, Telangana, India
V. Lakshmi Narasimhan, Elizabeth City State University, Elizabeth City, NC 27909, USA
D. Rammohan, CMR College of Engineering & Technology, Hyderabad/Bhagyanagar, Telangana, India
S. Kelkar, National Institute of Technology (NIT-K), Surathkal, India
|
| Paper Title |
Constructing an Immune-Related Gene Network of Host-Pathogen Interactions in Rice Plants Using Deep Learning Based Protocol
|
| Subject Category |
Computer Engineering
|
| Abstract |
This paper details the use of Deep Learning based protocol to construct an immune related Gene network of host-pathogen interactions in plants. Specifically, Deep Learning algorithms of LeCun, et.al. [1], Min, et.al. [2] and Zou, et.al. [3] have been used over three GEO datasets (GSE36272, GSE43050 and GSE95394) using the GPL2025 Affymetrix Rice Genome Array platform [9,10]. Our results confirm the following: i) a classification accuracy of 100% on test data using a three-layer deep neural network, ii) successful identification of top-ranked immune-related hub genes via Random Forest feature importance scoring, and iii) construction of statistically significant gene modules using the Markov Cluster Algorithm (MCL). When compared to related literature, the results presented in this paper are superior in the following terms: i) higher classification accuracy for host vs. pathogen-exposed conditions, ii) integration of multiple PPI and co-expression data sources with deep learning-based gene ranking, and iii) construction of rice-specific immune gene networks validated across multiple stress conditions and pathogen types.
|
| Keyword |
Deep Learning Protocol, Immune-Related Gene Network, Host-Pathogen Interactions, Rice Plants and Oryza sativa
|
|
| Paper ID |
IJIFR/V13/E8/003
|
| Author |
Reshmi R, Research Scholar, Research &PG Department of Commerce Government College Attingal,University Of Kerala Thiruvananthapuram
Prof. Dr. Antha S, Professor, Research Supervisor Research &PG Department of Commerce Government college for Women University of Kerala Thiruvananthapuram
|
| Paper Title |
TECH-DRIVEN TRANSFORMATION OF PARAMEDICAL ROLES: EMERGING SKILL SETS AND CHALLENGES
|
| Subject Category |
Commerce
|
| Abstract |
The role of paramedical professionals has drastically altered due to the increasing use of digital technologies in the healthcare settings. The workforce, who were only limited to assisting in medical roles, termed as paramedical professionals, are now expected to contribute as technologically competent members of modern healthcare systems. Besides presenting an overview of the new competencies expected of them, as well as the challenges faced in the process, this paper seeks to discuss the effects of digital technologies on the roles of paramedical professionals. The premise of this study is an overview of the existing research on workforce evolution and digital transformation in healthcare. According to the study, technologies such as telemedicine, artificial intelligence, the Internet of Things(IoT), and data-based healthcare systems are drastically altering role, functions and expectations, thus requiring skill development. Some of the challenges, however, include a lack of training opportunities, resistance to new technologies, and infrastructure challenges. The paper ends with an overview of how to improve the digital preparedness of paramedical professionals
|
| Keyword |
Paramedical Professionals, Digital Transformation, Healthcare Systems, Skill Development, Workforce Evolution
|
|
| Paper ID |
IJIFR/V13/E8/002
|
| Author |
Saranya B S, Sree Chitra Thirunal College of Engineering, Pappanamcode,Thiruvananthapuram, Kerala
Sidharth Prem, Sree Chitra Thirunal College of Engineering, Pappanamcode, Thiruvananthapuram, Kerala
Ashish, Sree Chitra Thirunal College of Engineering, Pappanamcode, Thiruvananthapuram, Kerala
N K Hanna Thasnim, Sree Chitra Thirunal College of Engineering, Pappanamcode, Thiruvananthapuram, Kerala
|
| Paper Title |
Data Cloud Implementation Over Data Processing
|
| Subject Category |
Computer Engineering
|
| Abstract |
Data-oriented methods dominate software development and data protection and data hiding Computer programs produce results by manipulating data. An important factor in determining the ease with which they can perform this task is how well the data types available in the language being used match the objects in the real-world of the problem. Therefore, a language supports an appropriate collection of data types and structures. In the earliest languages, all problem space data structures had to be modeled with only a few basic language-supported data structures linked lists, and binary trees were implemented with arrays. Cloud computing is a model that utilizes shared computing resources as a service over an Intranet or Internet to provide dynamic scalable computing or services. In cloud, common applications can access from a remote location and software applications run on the cloud servers and we can send data to the servers from user location. Users get a broad network to data access from a remote location and processing data in a location independent fashion. User can login from a remote location, start a session that requires search functions and data operations. Cloud computing provides on-demand network access to a shared pool of configurable computing resources. This paper concerns virtualization in cloud deployment model
|
| Keyword |
Virtualization, Mobile Cloud Computing, Service Level Agreements
|
|
| Paper ID |
IJIFR/V13/E8/001
|
| Author |
Dr Revathy Menon, Independent Researcher
|
| Paper Title |
Uncentering The Human: Robinson Jeffers And The Ethics Of Inhumanism
|
| Subject Category |
English
|
| Abstract |
Robinson Jeffers, a distinctive figure in twentieth-century American poetry, articulates a radical ecopoetic vision that challenges anthropocentric worldviews through his philosophy of ‘inhumanism’. His work reimagines the human-nature relationship by displacing the human as the central measure of value and emphasizing interdependence within a vast cosmic and geological continuum. Through vivid poetic landscapes shaped by the Californian coast, Jeffers critiques modern humanism and aligns with deep ecological principles, advocating an ethic of humility, reverence, and detachment. His vision foregrounds the autonomy and agency of the nonhuman world, crafting a poetics that is at once tragic, philosophical, and urgently relevant in the context of contemporary ecological crisis.
|
| Keyword |
: Inhumanism, Humanism, Ecopoetics, Anthropocentrism, Deep Ecology
|
|
|
|