Category Archives: Uncategorized

Authors: Prachi Durge, Mahek Shribas, Mohanish Lanjewar, Parth Gadwal, Pranay Wadibhasme, Pranjali Nakhate

Abstract: This paper presents a systematic comparative study of three tabular reinforcement learning (RL) algorithms—Q-Learning,State-Action-Reward-State- Action (SARSA), and Double Q-Learning—deployed within a configurable stochastic GridWorld environment. The environment incorporates slip-based stochastic transitions, trap cells, potential-based reward shaping grounded in the theoretical guarantees of Ng et al. [1], and partial observability modes. The central research hypothesis investigates whether Double Q-Learning’s decoupled selection-evaluation mechanism demonstrably reduces maximization bias compared to vanilla Q-Learning, particularly under elevated stochastic transition probabilities. An interactive web-based research platform is developed using Flask and Chart.js, enabling real-time policy visualization, value-function heatmaps, Q-table analysis, and multi-seed benchmark comparisons with confidence intervals. Experimental results across three canonical grid configurations demonstrate that Double Q- Learning achieves superior convergence stability and reduced overestimation in high-slip environments, while SARSA exhibits inherently conservative on-policy behavior that trades off peak performance for robustness near traps.

DOI: https://doi.org/10.5281/zenodo.19830361

Published by: vikaspatanker

Authors: Yashvi Rajiv Vyas, Mohammad Armaan, Rida Sadiqa, Sarayu Anand Gongada, Mohammed Sufyan, Dr. Chandrasekhar V (Project Giude)

Abstract: ExplorAR Glasses is an intelligent augmented reality (AR)-based travel assistance system designed to enhance real-world exploration through contextual digital augmentation. The system integrates geolocation, artificial intelligence, computer vision, and real-time API services to deliver immersive, hands-free assistance to users. By combining GPS-based location tracking, AI-generated contextual insights, OCR-based translation, weather forecasting, and voice interaction, ExplorAR enables users to interact with their surroundings in a seamless and intuitive manner. The system is built using a modular architecture consisting of a lightweight mobile client and a cloud-based backend. The backend leverages large language models (LLMs) for contextual information generation, while external APIs provide navigation, translation, and environmental data. The frontend prototype, developed using a cross-platform framework, serves as an intermediary between device sensors and backend services. The solution is designed to be scalable and adaptable for integration with wearable AR devices such as smart glasses. This project demonstrates how emerging technologies can be combined to create a real-time, context-aware digital assistant that improves accessibility, travel experience, and user interaction with physical environments.

DOI: https://doi.org/10.5281/zenodo.19828864

Published by: vikaspatanker

Authors: Arayan Gandre, Swaraj Sakpal, Unmesh Nhavelkar, Vedant Gaikwad, Prof. Smita Pawar

Abstract: In today’s highly competitive and data- driven marketplace, pricing strategy has become a decisive factor in determining a company’s profitability, customer satisfaction, and long-term sustainability. Traditional static pricing models, which rely on fixed markups or manually updated price lists, are often inadequate in responding to the dynamic nature of modern markets. These methods struggle to adapt to frequent fluctuations in consumer demand, competitor actions, supply chain disruptions, and seasonal influences. This research presents the design and development of an Artificial Intelligence (AI)-based Dynamic Pricing Engine that autonomously predicts and optimizes product prices in real time. The proposed framework integrates a variety of heterogeneous data sources — including historical sales transactions, customer purchasing behavior, inventory levels, market demand elasticity, and competitor pricing trends — to generate context-aware pricing recommendations. The system employs a hybrid machine learning approach: regression-based models are used for short- term price prediction, while reinforcement learning techniques enable continuous self-improvement through feedback-driven optimization. A prototype implementation was tested using real-world re- tail and e-commerce datasets to evaluate its effectiveness. The experimental results demonstrate that the AI-driven dynamic pricing model significantly enhances revenue optimization, profit margins, and inventory turnover compared to traditional rule- based or static pricing systems. Moreover, the model exhibits rapid adaptability to demand shifts and improved decision- making accuracy under volatile market conditions. The findings highlight the transformative potential of AI in automating strategic business decisions and emphasize the scalability and robustness of intelligent pricing systems. This study contributes to the broader field of intelligent commerce by providing a data-centric, adaptive, and scalable solution for modern enterprises seeking to maintain competitiveness in the evolving digital economy.

Published by: vikaspatanker

Authors: Sareddy Prasanna Reddy, Dr. P. Kowstubha, Parameshwari Rathod, Barla Ananda Sagar

Abstract: This paper presents the design and implementation of DC-DC buck converter using a digital PI control technique. The system converts a 24V DC input into a regulated 12V DC output. An microcontroller is used to implement closed-loop control and generate PWM signals. The controller continuously monitors the output voltage and adjusts the duty cycle to maintain stable output under different load conditions. The converter achieves an efficiency of around 90% with good voltage regulation. The results show that the proposed system is suitable for low-power embedded applications.

Published by: vikaspatanker

Authors: Ms.C. Harivarshini, Ms.M. Shubhashree, Dr.R. Karthik

Abstract: The global workforce is undergoing rapid evolution. Current data-driven research into worldwide employment trends thus has become a pressing need. The objective of the present study was to conduct a comprehensive analysis of job advertisements and salary trends by reviewing 999 job records collected from 213 countries; these included a total of 13 data points. As part of its analytic process, the present study utilized a data preprocessing pipeline that involved the passing of data through multiple stages – data cleansing, data type conversion, aggregation, data partitioning, normalization, etc., prior to submission to various data visualisation techniques; these included bar charts, histograms, box plots, scatter plots, correlation heat maps, skills frequency heat maps, pie charts, violin plots, and choropleth maps. Among the most significant findings of the present study were the following: job advertisements show evidence of consistent salary levels based on both level of education and type of job; however, geographic region and industry sector appear to play an important role in determining salary levels. Additionally, the study concluded that the most common skills necessary for attaining such salaries are as follows: management skills; analytical skills; design skills; communication skills; and technical/data oriented skills. Consequently, the authors propose a framework that can be used to better understand the trends of employment and provide actionable insights into the employment market.

DOI: https://doi.org/10.5281/zenodo.19816389

Published by: vikaspatanker

Authors: P Sudhakar Reddy, Pothireddy Nithisha, Sakamuru Hari priya, Saggam Ranjith Kumar, Panditi Prem Kumar, Yagnam setty chaithanya kumar

Abstract: Increasing water pollution due to floating solid waste in rivers, lakes, and drainage canals has become a major environmental concern, and manual waste collection in water bodies is inefficient, unsafe, and time-consuming. This project presents Roboclean, an ESP32-based automated garbage collection system designed specifically for collecting floating waste from water surfaces using a conveyor belt mechanism. The system employs dual conveyor belts driven by DC motors through motor driver modules to lift and transfer waste from water to a collection bin. An ESP32 microcontroller acts as the central control unit, coordinating motor operations and system monitoring. IoT connectivity using the Blynk platform enables real-time remote control and monitoring of the system through a mobile application. A 16×2 LCD display provides on-site status information, while a regulated power supply ensures reliable operation. By automating floating waste collection and enabling remote supervision, the proposed system reduces manual labor, improves safety, and enhances cleanliness of water bodies. Roboclean offers a cost-effective and scalable solution suitable for rivers, lakes, sewage canals, and smart city environmental management applications.

DOI:

Published by: vikaspatanker

Authors: Kura Sairam, Kurva Saisharath, Dr. P. Kowstubha, A. Sai Aditya

Abstract: Renewable energy sources such as solar power are highly dependent on environmental conditions, which often leads to fluctuations in output voltage and current. These variations create challenges for conventional inverter systems like Voltage Source Inverters (VSI), Current Source Inverters (CSI), and even traditional Z-Source Inverters (ZSI), affecting their efficiency and reliability. To address these issues, this paper focuses on the design and simulation of a Quasi Z-Source Inverter (QZSI) for photovoltaic (PV) energy conversion. The QZSI is an improved version of the ZSI, achieved by modifying the impedance network. This topology offers several advantages, including the ability to perform both buck and boost operations in a single stage, reduced component stress, and a continuous input current, which is particularly beneficial for PV systems. Additionally, the QZSI allows the use of shoot- through states without damaging the inverter, enabling effective voltage boosting under varying input conditions. In this work, the operating principle, voltage boost capability, and control strategy of the QZSI are studied. A simulation model is developed using MATLAB/Simulink to evaluate system performance under different operating scenarios. The results demonstrate that the QZSI provides improved voltage stability and overall efficiency, making it a suitable choice for renewable energy applications.

DOI: https://doi.org/10.5281/zenodo.19766037

Published by: vikaspatanker

Authors: Dr. Jonathan Miller, Dr. Emily Carter, Michael Anderson, Dr. Sophia Reynolds, Daniel Thompson, Chaitanya Srinivas

Abstract: The rapid evolution of cloud computing has significantly transformed the role of database leadership, necessitating a shift from traditional management approaches to dynamic, automation-driven, and resilience-oriented strategies. This paper explores the redefinition of database leadership within cloud-native environments, where scalability, distributed architectures, and continuous integration and deployment pipelines are essential. It highlights the importance of leveraging automation, intelligent monitoring, and self-healing systems to ensure high availability and operational resilience. The study addresses key challenges such as maintaining data consistency across distributed systems, ensuring security in multi-tenant cloud environments, and optimizing performance under variable workloads. Furthermore, it examines how modern leadership practices incorporate cloud-native principles, including microservices architecture, containerization, and Infrastructure as Code (IaC), to enhance efficiency and system reliability. Based on conceptual analysis and practical insights, the paper proposes a strategic framework that emphasizes proactive decision-making, automation adoption, and resilience engineering to achieve scalable, fault-tolerant, and robust database systems while minimizing operational risks and downtime, ultimately underscoring the critical role of adaptive leadership in meeting the demands of modern cloud-native ecosystems.

DOI: https://doi.org/10.5281/zenodo.19765416

Published by: vikaspatanker

Authors: Dr. Jonathan Reed, Dr. Emily Carter, Michael Thompson, Dr. Sarah Williams, David Anderson, Chaitanya Srinivas

Abstract: Modern enterprise platforms increasingly depend on data from multiple heterogeneous sources such as legacy systems, cloud applications, and real-time streams, making scalable and efficient data integration a critical challenge. This paper presents a comprehensive study of data integration architectures for multi-source enterprise environments, with a particular focus on Extract, Transform, Load (ETL) processes and Oracle Data Integrator (ODI) implementations. It evaluates centralized, distributed, and hybrid architectural models to determine their effectiveness in handling large-scale and high-velocity data workloads. An empirical analysis based on real-world enterprise scenarios is conducted to assess key performance factors including scalability, data consistency, fault tolerance, and maintainability. The study further investigates the role of ETL pipelines in enabling structured data transformation and highlights how ODI’s declarative approach and pushdown optimization techniques improve processing efficiency. Additionally, best practices such as parallel processing, metadata-driven integration, and incremental data loading are explored to enhance system performance. The results demonstrate that the integration of robust ETL strategies with ODI-based optimizations significantly improves throughput and reduces latency in complex enterprise systems, providing valuable insights for designing scalable and reliable data integration solutions.

DOI: https://doi.org/10.5281/zenodo.19765226

Published by: vikaspatanker

Authors: Dr. James Anderson, Emily Carter, Dr. Michael Thompson, Daniel Roberts, Dr. Sophia Williams, Chaitanya Srinivas

Abstract: The rapid growth of data-intensive applications has necessitated the adoption of diverse data storage technologies to meet evolving performance, scalability, and reliability requirements. Traditional single-database approaches often fail to address the heterogeneous data needs of modern systems, leading to inefficiencies in data management and processing. This research proposes a unified hybrid persistence framework that integrates in-memory, NoSQL, and relational databases—specifically Redis, MongoDB, and PostgreSQL—to optimize data storage and retrieval strategies in high-performance environments. The framework leverages Redis for low-latency caching and real-time data access, MongoDB for flexible schema design and efficient handling of semi-structured data, and PostgreSQL for strong transactional integrity and advanced querying capabilities. By combining these systems within a cohesive architecture, the proposed approach enables intelligent data tiering, workload distribution, and consistency management. Furthermore, the study introduces adaptive data routing and synchronization mechanisms to ensure seamless interoperability across multiple persistence layers. Experimental evaluation indicates that the proposed framework significantly improves system throughput, reduces query response time, and enhances scalability compared to traditional monolithic database solutions. Additionally, it strengthens fault tolerance and supports dynamic scaling in distributed environments, making it highly suitable for modern cloud-native and enterprise-scale applications.

DOI: https://doi.org/10.5281/zenodo.19763840

Published by: vikaspatanker