Handwritten Digit Recognition on the MNIST Dataset using Convolutional Neural Networks
Information Technology Usage and Supply Chain Performance: A Two-Stage Disjoint Structural Equation Framework of Sequential Mediation through Visibility and Resilience
AI-Powered Real-Time Stock Price Estimator
Fake News Detection using Machine Learning
Benchmarking of NoSQL databases
A Study on Spending Patterns in the Digital Era with Special Reference to Tamilnadu
Enhancing Donor Acquisition and Retention in Blood Banks via AI-Powered Decision Support Framework
Computer-Based Fuzzy Logic for Forecasting the Population Census of Edo State, Nigeria
Artificial Intelligence in Investment Management, Asset Management and Warehouse Management
Influence of Digital Transformation and Artificial Intelligence in Business
A Comparative Analysis for Identifying the Polarity of People Based on Emotional Pulse in a Smart City
Video Analytics for Optimizing Bank Services
AI-Powered Chatbots for Customer Service: Effectiveness and User Satisfaction
Role of Artificial Intelligence in Investment Management
Fake News Detection using Machine Learning
Handwritten digit recognition helps make data entry quicker in places like banking and healthcare. The MNIST dataset includes 70,000 black and white images of numbers from 0 to 9, and it is frequently used to check how well different models can classify things. In this study, Convolutional Neural Networks (CNNs) were used, which are great at learning patterns from images on their own, so they need less help from people when getting the data ready. After resizing all the images to the same size and turning the digit numbers into a special format, they trained the proposed CNN models. These models achieved over 99% accuracy, which is better than older techniques like Support Vector Machines (SVM) and Multilayer Perceptrons (MLP). These results show that CNNs are dependable and work well in real-world situations. Newer approaches, such as combining quantum and classical networks or using ensemble models, are also being explored for improved performance.
Supply Chain Management (SCM) involves the coordinated management of procurement, transformation, and distribution activities, including demand forecasting, production planning, inventory control, logistics, and customer relationship management. Although information technology (IT) has enhanced supply chain capabilities, empirical evidence on its influence on supply chain visibility (SCV), resilience (SCR), and performance (SCP) remains limited in emerging economies, with existing studies largely fragmented. While prior research has examined direct relationships among these constructs, the sequential mediating roles of SCV and SCR between IT usage and SCP require further exploration. This study adopts a quantitative, cross-sectional design using data from 600 supply chain managers from BSE- and NSE-listed firms in India, collected through a two-stage cluster sampling approach. Validated measurement scales were assessed for reliability and validity using confirmatory factor analysis, and hypothesized relationships were tested using PLS-SEM version 4. A two-stage disjoint SEM framework was applied to analyze how IT components collectively influence supply chain outcomes through layered capability development, with results confirming that IT usage positively affects SCV, SCR, and SCP. Furthermore, SCV and SCR exhibit significant sequential mediation between IT usage and SCP, supporting the Resource-Based View and Dynamic Capabilities View.
The financial market is dynamic and highly volatile, making accurate and timely stock price prediction a challenging yet valuable task. This study presents an AI-Powered Real-Time Stock Price Estimator that leverages machine learning techniques to forecast stock prices based on live market data. The system integrates real-time data collection from financial APIs with predictive models such as LSTM (Long Short-Term Memory) networks, which are well-suited for time series forecasting. It processes historical stock data along with live feeds to continuously update predictions and provide users with near-instant insights into future price movements. The model is trained and evaluated using a range of performance metrics to ensure accuracy and responsiveness. This solution aims to assist traders, investors, and financial analysts in making informed decisions by combining the power of artificial intelligence with real-time data analysis. The initiative demonstrates the potential of AI in transforming traditional stock market forecasting into a more dynamic and adaptive process.
In today's digital age, social media platforms have become breeding grounds for fake news, false information disguised as legitimate news. This misinformation spreads faster than ever before, influencing public opinion, creating social unrest, and even affecting election outcomes. While fact-checkers work to verify news manually, they can't possibly keep up with the millions of posts shared every minute. The challenge lies in developing automated systems that can quickly and accurately identify fake news at scale. This research aims to develop and evaluate machine learning models that can automatically detect fake news by analyzing linguistic patterns in the text, source reliability indicators, social media engagement metrics, and cross-referencing with verified information sources. The study includes the comparison of different machine learning approaches (including traditional classifiers and deep learning models) to determine which methods work best for different types of misinformation. The findings of this research paper demonstrate that machine learning can effectively identify fake news with high accuracy, especially when combining multiple detection approaches. However, challenges remain in keeping up with constantly evolving misinformation tactics. Future systems will need to incorporate real-time learning capabilities and explainable AI techniques to maintain effectiveness and user trust. These automated detection tools show great promise in helping create a more truthful online information environment.
As more data-intensive applications emerge, higher-performance, scalable, and domain-specific databases have become increasingly needed. In this paper, comparative benchmarking of three leading NoSQL databases, Redis, AstraDB, and Neo4j, each offering key-value, document/wide-column, and graph data models, respectively, is presented. The benchmarks contrast the performance of each database in performing basic CRUD operations by measuring execution time, CPU and memory consumption, disk I/O, and network throughput. In the experiment, a Python-based test framework is used that captures real-time system statistics based on the psutil library. The experiments reveal considerable performance trade-offs between the databases, with Redis being the most performance-focused and requiring the least memory, AstraDB offering document flexibility, and Neo4j offering superior performance in performing complicated relationship queries. This research facilitates informed database choice based on application-specific needs.
