Recent Trends in Artificial Intelligence & It’s Applications (e-ISSN: 2583-4819)

AI in the Everyday: Machine Learning’s Role in Reshaping Digital Culture and Practices

Ismail Olaniyi MURAINA, Bashir Oyeniran Ayinde, Tajudeen Mohammed — Mon, 02 Mar 2026 00:00:00 +0000

This article examines how embedded Machine Learning (ML) systems are reshaping everyday digital experiences, including social media use, smart-enabled devices, and recommendation-driven consumption. Drawing on recent empirical studies and policy reports, the article synthesizes secondary evidence indicating the rapid expansion of generative AI and large language model–based platforms, with current global estimates suggesting over one billion monthly active users. At the infrastructural level, industry analyses consistently report that a substantial proportion of connected devices now incorporate some form of ML functionality, highlighting the growing pervasiveness of AI in routine digital life. Social media platforms, serving approximately 5.66 billion global user accounts, are increasingly organized around algorithmic curation, personalization, and automated content moderation, reshaping patterns of visibility, attention, and engagement within digital cultures. Building on interdisciplinary research, this review examines how opaque algorithmic decision-making influences user perceptions, creative practices, and social interaction, while simultaneously transforming expectations related to convenience, privacy, and trust. Secondary survey evidence further points to a contextual divide in public attitudes toward AI technologies. While a majority of adults in the United States report comfort with AI-based recommendation systems, acceptance declines notably when similar technologies are associated with profiling, facial recognition, or surveillance-oriented applications. Overall, the reviewed literature suggests that artificial intelligence is not merely augmenting existing digital cultures, but is actively reconfiguring social practices, power relations, and cultural imaginaries within data-driven environments.

JoyNest: An Intelligent Web-Based Event Planning and Vendor Recommendation System

Thu, 12 Mar 2026 00:00:00 +0000

In today’s fast-paced lifestyle, organizing events such as technical programs, birthdays, weddings, and social gatherings has become increasingly challenging. Poor planning, lack of coordination, budget mismanagement, and limited communication between organizers and vendors often result in confusion, higher expenses, and reduced satisfaction. To address these issues, this paper introduces JoyNest, a web-based event management application designed to simplify and streamline the entire event planning process. JoyNest provides users with a structured and user-friendly platform where they can plan, organize, and monitor their events efficiently. The system guides users step-by-step, beginning from event selection to vendor management and invitation handling. An integrated AI-based assistant supports users by offering smart suggestions, reminders, and personalized recommendations based on their preferences and requirements. The platform also connects users with registered vendors, enabling smooth communication and service selection within a single system. By integrating frontend technologies like React.js and backend processing using Python, the application ensures responsive performance, secure data handling, and real-time interaction. JoyNest not only improves event organization and user satisfaction but also creates business opportunities for vendors by promoting their services on a unified digital platform. Overall, the proposed system aims to make event planning more organized, cost-effective, and convenient through the use of modern web technologies and intelligent assistance.

Explainable AI-Based Intelligent Closed-Loop Drug Delivery System for Mean Arterial Blood Pressure (MABP) Using Controlled Drug Administration

N. B. Mahesh Kumar — Wed, 18 Mar 2026 00:00:00 +0000

Mean Arterial Blood Pressure (MABP) regulation using vasoactive drugs remains challenging in intensive care units due to nonlinear patient dynamics, inter-patient variability, and limitations of manual titration protocols. This study introduces an ensemble closed-loop controller that integrates Proportional-Integral-Derivative (PID), Fuzzy Logic, Model Predictive Control (MPC), and Reinforcement Learning (RL) algorithms, augmented by SHapley Additive exPlanations (SHAP)-based Explainable AI (XAI) for real-time arterial signal processing and interpretable decision-making. The system employs a physiologically validated nonlinear patient model simulating MABP dynamics under realistic Gaussian noise conditions. Reinforcement Learning demonstrates superior setpoint tracking and settling performance, while MPC provides strong foresight-based optimization; PID and Fuzzy Logic offer robust classical and expert-knowledge-driven baselines. An ensemble fusion strategy weights contributions from each algorithm to leverage complementary strengths, addressing individual limitations like computational intensity or data requirements. XAI integration delivers real-time feature attributions highlighting trajectory lag and error as dominant drivers building clinician trust for regulatory approval. PySpark scalability supports multi-patient validation, with a structured pathway from synthetic cohorts to randomized controlled trials, promising enhanced hemodynamic stability and reduced nursing workload in clinical deployment.

A Multimodal Deep Learning Framework for Predicting Social Demeanor through Sentiment Analysis of Social Media Videos

Mritunjay Kr. Ranjan, S. K. Tiwari, Ankita Patil — Wed, 18 Mar 2026 00:00:00 +0000

Social media videos capture abundant social and emotional interactions that traditional sentiment analysis does not focus on, as it usually focuses on text or image. This paper introduces a novel model of predicting social demeanor on the basis of video-based sentiment analysis on social media. It uses EfficientNet in deep visual feature extraction and DeepFace in rich facial emotion recognition. They are inputted into a multi-output neural network which categorizes both coarse sentiments (Positive, Negative, Neutral) and subtle social demeanors (Friendly, Supportive, Respectful, Hostile, Dismissive, Rude, Neutral) together. The model is trained and tested with a balanced dataset of over 3000 videos and exceeding 60000 manually extracted frame and an in-depth sentiment and social demeanor annotations. Strong metrics are used to evaluate the system i.e. precision, recall, F1-score, and ROC-AUC. Experimenting with real-world videos confirms the system to be working as far as its ability to identify small social signals is concerned, making it possible to develop an in-depth analysis of social interactions over the internet. The project fills in the sentiment analysis and social behavior modeling gap with a cognitive science-based hierarchical methodology, providing a scalable, interpretative, and ethical way of social media analysis, content moderation, and mental health monitoring.

Privacy Risk Detection System for Social Media

Dipti Patil, Vaibhavee Valanju, Varad Anjankar, Aayush Asawale — Wed, 01 Apr 2026 00:00:00 +0000

Nowadays, Social media has become an essential part of daily life, as it allows individuals to share their ideas, experiences, and personal movements with all users; however, it sometimes poses privacy risks or threats by revealing personal information. Users usually post images, videos, and text without realizing that this might reveal personal or sensitive information, such as location, Personal Identification Numbers, or details that could be exploited. This project will help to present an AI powered Application that will detect and alert users about privacy risk that will be caused by sharing posts on social media. This system will use the concept of Deep Learning, Natural Language Processing (NLP), and also Image Processing techniques to analyze both visual and textual data. It scans uploaded media for features like faces, license plates, geotags, documents, or background elements that may expose personal data. It also analyses captions and comments for mentions of sensitive information such as phone numbers, addresses, or plans that may attract malicious intent. The latest NLP and deep learning methods, like Named Entity Recognition, transformer-based detection, stylometry, topic modelling, privacy-preserving learning, Autoencoders, CNNs (Convolutional Neural Networks), LSTM- based Anomaly Detection, and AI-Powered Image Anonymization, are used. After the risk is detected, the scanner gives the real-time suggestions or warnings before the post is uploaded. This tool is designed to be user-friendly and works by allowing seamless privacy checks without interrupting the user experience. The main objective of this project is to encourage responsible sharing by making users more aware of the consequences of posts. It serves as a proactive approach to digital privacy, aiming to reduce cases of identity theft, stalking, and data misuse on social media platforms. With increasing threats to personal data, this solution brings much- needed attention to privacy awareness in the online world.

Early Detection of Fake News in Social Media: A Hybrid Multimodal Architecture

Suraj Pal, Devender Kumar — Wed, 01 Apr 2026 00:00:00 +0000

The rapid proliferation of fake news on social media platforms has emerged as a critical societal challenge, undermining public trust, distorting political discourse, and impacting real-world decision-making. This paper presents a comprehensive hybrid multimodal framework for the early detection of fake news, designed to operate in the initial stages of news propagation — within the first 5 to 15 minutes of a news post appearing online. The proposed architecture integrates three complementary streams of information: (1) news content features extracted using both GloVe-CNN and transformer-based (RoBERTa) models, (2) social context features derived from tweet responses and user credibility profiles, and (3) visual features from accompanying images processed through a ResNet-50 backbone. These streams are fused using a transformer-style cross-modal attention mechanism, enabling the model to capture complex inter-modal dependencies. The social context module employs a self-attention mechanism to identify discriminative tweets globally, and Gated Recurrent Units (GRUs) to model temporal dynamics locally. They evaluate the proposed framework on the FakeNewsNet dataset (PolitiFact and GossipCop subsets) as well as Fakeddit, Twitter15, and Weibo benchmarks. The proposed model achieves an accuracy of 89.5% on Fakeddit, 93.1% on Twitter15, 88.3% on Weibo, and 87.8% on FakeNewsNet — consistently outperforming existing state-of-the-art models, including Att-RNN, EANN, MVAE, and BMMFN. Ablation studies confirm the contribution of each module, and explainability analyses using Grad-CAM and SHAP validate the interpretability of the model's decisions.

An Intelligent Facial Recognition-Based Automated Student Attendance System with Schedule Integration and Absentee Notification for Educational Institutions

Nadhiya S, I. Ajitha — Wed, 08 Apr 2026 00:00:00 +0000

Traditional methods for tracking attendance in schools take a lot of time, have many errors, and can lead to student impersonation for attendance purposes (called proxy attendance). To overcome these issues, they have developed an automated attendance management system that uses face recognition technology. The system uses the Haar Cascade Classifier to detect faces and the OpenCV Template Matching algorithm to recognize them. The overall application was created in Python with the Flask web framework and includes a full-featured SQLite database, an authentication system that is based on each user's role and requires validation of attendance against a student's schedule before attending class, and provides automated notifications via email regarding attendance changes. Using the new system, the time required to take attendance for a class of 60 students dropped from 12-15 minutes to 45-60 seconds (92% time savings). Precision (actual recorded attendance for students) was 97.8%; recall (student's face actually being recorded) was 96.5%; F1-score (measure of accuracy) was 97.1%. In conclusion, this will be an effective, secure, and scalable solution for attendance management in educational institutions.

From Cloud to Edge: Advancing Real-Time AI Applications

Wed, 08 Apr 2026 00:00:00 +0000

Edge Computing together with Real-Time Artificial Intelligence (AI) is gradually changing how modern computing system works. Instead of relying completely on centralized cloud servers, this approach focuses on processing close to where it is created. In simple terms, devices like sensors, smart devices, and embedded systems can analyse data directly on the device. This reduces communication delay, lowers bandwidth usage, and helps systems respond more quickly. When methods like Machine Learning and Deep Learning are used in these edge devices, they can analyze data in real time and make fast decisions without waiting for cloud support. The rapid growth of the Internet of Things has resulted in a large volume of real-time data being generated every second. Sending all this data to distant cloud servers is not always efficient, especially in situations where time is critical. For example, in autonomous vehicles or industrial automation systems, even a slight delay can affect performance or safety. Because of this, edge AI becomes important. Local data processing also improves privacy since sensitive information does not always need to travel across networks. At the same time, cloud platforms still play an important role in training complex AI models and storing large datasets, while edge devices mainly handle real-time predictions. Recent improvements in hardware technology, lightweight AI models, and energy-efficient processors have made it easier to deploy AI at the edge. These technological advancements improve scalability and reliability in distributed systems. Overall, the combination of edge computing and AI supports distributed systems that can provide faster and more secure services in many industries. As technology advances, this integration will become even more important in supporting future real-time applications.

Artificial Intelligence in Autonomous Systems: Applications, Challenges, and Future Directions

Suraj R. Nalawade, Tapase H. O, Aryan Beloshe — Wed, 08 Apr 2026 00:00:00 +0000

Autonomous systems, driven by rapid advancements in Artificial Intelligence (AI), are transforming industries by enabling machines to work on their own, make decisions, and engage with their surroundings without needing human input. This paper delves into the growth, applications, challenges, and future trends of autonomous systems, highlighting the crucial role of AI technologies like machine learning, computer vision, and sensor integration. It looks at how these systems are reshaping sectors such as transportation (e.g., self-driving cars), logistics, manufacturing, and healthcare, by enhancing efficiency, safety, and scalability. However, despite their immense potential, these systems still face major hurdles, such as reliability issues, ethical dilemmas, regulatory concerns, and the need to build public trust. The paper also covers the evolving development of AI algorithms and sensor technologies, and the importance of interdisciplinary collaboration to tackle these challenges. Through a thorough examination of both the opportunities and limitations of autonomous systems, this research aims to provide valuable insights into the future of intelligent, self-operating machines and their impact on society.

ABC-SVM: Artificial Bee Colony Optimization for Feature Selection in Medical Diagnosis: A Comprehensive Literature Survey

Satish Kumar Kalagotla, Thoudam Basanta, Mutum Bidyarani Devi — Wed, 08 Apr 2026 00:00:00 +0000

Feature selection is a critical preprocessing step in machine learning, particularly for medical diagnosis, where high-dimensional datasets often contain irrelevant or redundant features that degrade classifier performance. The Artificial Bee Colony algorithm, inspired by the intelligent foraging behavior of honey bees, has emerged as a powerful metaheuristic for solving complex feature selection problems. This literature survey provides a comprehensive examination of Artificial Bee Colony-based feature selection frameworks integrated with Support Vector Machines for medical diagnosis. The paper reviews the theoretical foundations of the Artificial Bee Colony algorithm, its biological inspiration, and operational phases involving employed, onlooker, and scout bees. Various representation schemes for feature selection, including binary encoding and multi-objective fitness functions, are analyzed. The survey critically evaluates empirical results from comparative studies against Genetic Algorithms and Particle Swarm Optimization across multiple medical datasets. Key findings reveal that the Artificial Bee Colony with Support Vector Machine achieves substantial feature reduction, ranging from 40% to 64%, while maintaining classification accuracy between 88% and 98%. Research gaps are identified, including the need for adaptive parameter control, handling of high-dimensional data, and integration with deep learning architectures. The survey concludes that Artificial Bee Colony with Support Vector Machine represents a promising direction for developing parsimonious, accurate, and interpretable medical diagnosis systems.