Journal of Artificial intelligence and Machine Learning

Synergistic Synthesis-Guided Parallel Incremental Optimization (SSPIO): A Unified Framework for Recursive, Comparative, and Nested Aggregate Query Processing

Rama chandra Reddy Vangala

Modern analytical workloads increasingly rely on recursive computations, correlated nested aggregates, comparison-heavy conjunctive queries, and massively parallel joins. Traditional query optimizers, which focus primarily on loop-free relational algebra, struggle to address the combinatorial explosion of recursive evaluation, join ordering, and incremental maintenance. In this paper, we propose a novel framework called Synergistic Synthesis-Guided Parallel Incremental Optimization (SSPIO), which integrates program synthesis, constraint-guided rewrite discovery, acyclic comparison evaluation, parallel join enumeration, and incremental aggregate indexing into a unified optimization architecture. SSPIO introduces a new meta-optimization paradigm that combines synthesis-based recursive rewriting with constraint-aware equivalence verification and parallel search-space pruning. Furthermore, we propose a novel data structure, the Comparative Recursive Partial Aggregate Graph (CRPAG), which generalizes partial aggregate indexes and acyclic comparison structures for recursive and correlated nested queries. Experimental evaluation demonstrates substantial performance improvements in recursive Datalog programs, SQL workloads with nested aggregates, and large multi-join analytical queries. SSPIO achieves up to three orders of magnitude speedup compared to baseline systems while maintaining provable correctness guarantees.

Feature Serving at Scale: Design, Migration, and Empirical Evaluation of a Production Feature Platform

Karthik Perikala

Modern machine learning systems depend on large, continuously evolving feature sets to support both offlinetrainingandlow- latencyonlineinference. As feature ecosystems grow, bespoke pipelines and tightly coupled application logic becomeincreas- ingly difficult to operate, leading to inconsistent fea- ture definitions, operational fragility, and slow iter- ation cycles. This paper presents the design and evolution of a production feature serving platform, beginning with an in-house NoSQL- based Feature Mart and fol- lowed by a migration to a managed Feature Store. Thein-housesystemprovidedflexibleschemadesign and low-latency access but required substantial en- gineering effort to maintain ingestion pipelines, fea- ture registries, refresh workflows, and serving mi- croservices. The managed platform introduced automated in- gestion from analytical sources, built-in feature gov- ernance, and native serving interfaces, fundamen- tally shifting the operational model of feature lifecy- cle management.Using normalized empirical mea- surements, we evaluate latency percentiles, through- put stability, and execution behavior under live con- figuration updates. Our results show that managed feature stores can preserve predictable serving performance while sig- nificantly reducing operational complexity, making them a compelling foundation for large-scale pro- duction machine learning systems. Keywords: feature serving, feature store, ma- chine learning systems, NoSQL, production ML, low-latency inference

Decentralized Trust: Consensus Mechanisms and Blockchain Integration In Distributed Systems

Divya Soundarapandian

This research investigates the application of the Grey Relational Analysis (GRA) method to evaluate and rank various decentralized computing alternatives, including Decentralized Systems, Networked Systems, Distributed Computing Architectures, Multi-node Systems, and Federated Systems. These systems play a vital role in modern digital infrastructure by enhancing performance, reducing bottlenecks, and improving security. To determine the most suitable alternative, the study considers key evaluation parameters such as availability, response time, cost, speed, and storage capacity. The GRA methodology provides a robust multi-criteria decision-making framework, enabling objective comparison and selection among the alternatives. Research Significance: With the exponential growth in data generation and processing needs, centralized systems face challenges in scalability, reliability, and latency. Decentralized and distributed architectures offer viable solutions, supporting resilient and efficient computing environments. This research is significant as it applies a structured decision-making approach to compare these emerging architectures, helping stakeholders adopt the most suitable system for their needs based on performance and cost-efficiency. Methodology: Grey Relational Analysis (GRA) GRA is a powerful multi-criteria decision-making tool used to handle complex systems with incomplete or uncertain data. In this study, GRA is used to evaluate alternatives against normalized criteria values. The method involves calculating the grey relational coefficients and grades to rank the alternatives. GRA’s ability to accommodate varying units and scales of parameters makes it particularly useful in technology evaluations. Alternatives: Decentralized Systems, Networked Systems, Distributed Computing Architectures, Multi-node Systems, Federated Systems. Each of these architectures is assessed for its ability to support distributed computing environments with robust performance and operational flexibility. Evaluation Parameters: Availability: Uptime and reliability of the system, Response Time: Speed of the system, n reacting to user or node requests. Cost: Overall implementation and maintenance expenses. Speed: Data processing and communication efficiency. Storage Capacity: Ability to manage and store large volumes of data. Result: Based on the GRA analysis, Distributed Computing Architectures emerged as the most suitable alternative, offering balanced performance across all evaluation parameters. Federated Systems and Multi-node Systems followed closely, particularly excelling in response time and availability. The results highlight the practicality of GRA in systematically evaluating technological choices, helping organizations make data-driven decisions. Keywords: Decentralized Systems, Grey Relational Analysis, Distributed Computing, System Evaluation, Multi-Criteria Decision Making, Federated Architecture

CFD Analysis of Artificial Lift Down hole tool Separator

Ramamurthy Narasimhan

This white paper evaluates the capability of commercially available Computational Fluid Dynamics (CFD) software to predict the performanceof an artificial lift downhole sand and gas separator. Theneed for a rigorous numerical andparametricanalysis originated fromtheproductmanagement team’s objective to strengthen the design basis for a market-leading separator. Existing empirical correlations have limitations, and the theoretical investigation originally envisioned was not sufficient to represent the complex multiphase flow physics inside the tool. Therefore, the product team selected CFD as a practical method to simulate separation behavior, infer important parameter relationships, and support future optimization.
A second business objective of this CFD analysis is to provide a stronger technical basis for communicating separator performance to customers. Historically, the product team and sales organization have found it difficult to explain the behavior of complex downhole separation physics using only empirical evidence. CFD simulation provides a visual and quantitative method to demonstrate how the separator functions and to highlight the technical differentiation of the product. Once the numerical model is validated through physical testing, it can reduce dependence on repeated prototype testing, which is costly and timeconsuming. A validated multiphase model can also support sensitivity studies, parametric optimization, and future product development. This white paper therefore establishes a virtual test-bench methodology for evaluating flow behavior, visualizing separation mechanisms, and estimating the performance of an artificial lift downhole separator.

Real-Time Predictive Optimization Framework for Multi-Platform Digital Advertising Using Machine Learning and Multi-Criteria Decision Analysis

Gaurav Saxena

The rapid growth of digital advertising across search engines, social media platforms, video-sharing services, and mobile applications has significantly increased the complexity of campaign management and advertisement optimization. Modern advertising ecosystems generate large volumes of real-time behavioral data, requiring intelligent decision-support mechanisms capable of continuously adapting advertisement placement, budget allocation, and platform selection. Conventional optimization approaches predominantly rely on static bidding strategies or rule-based decision models that often fail to respond effectively to rapidly changing user behavior and dynamic market conditions. Although machine learning has demonstrated considerable success in predicting campaign performance, existing studies rarely integrate predictive analytics with multi-criteria decisionmaking to support real-time advertising optimization across multiple digital platforms. This paper proposes an AI-Driven Real-Time Predictive Optimization Framework (AIPOF) that integrates machine learningbased performance prediction with Multi-Criteria Decision Analysis (MCDA) to improve multi-platform digital advertising decisions. The proposed framework employs predictive analytics to estimate campaign performance using key advertising indicators, including targeting accuracy, real-time responsiveness, platform compatibility, and cost efficiency. These predicted performance measures are subsequently incorporated into a Complex Proportional Assessment (COPRAS) model to rank alternative advertising strategies and identify the most effective campaign configuration under multiple competing objectives. The proposed framework is evaluated through a comparative analysis of five representative digital advertising strategies using standardized performance criteria and objective weighting mechanisms. Experimental results demonstrate that integrating predictive analytics with multi-criteria decision analysis enables more balanced and informed advertising decisions by simultaneously considering campaign effectiveness, computational responsiveness, cross-platform adaptability, and advertising cost. The comparative evaluation further indicates that AI-assisted optimization provides superior decision consistency and supports efficient resource allocation in dynamic advertising environments. The proposed AIPOF framework contributes to intelligent digital marketing by bridging the gap between predictive machine learning and decision optimization within real-time advertising ecosystems. The framework offers a practical decision-support solution that can assist marketing professionals, advertising platforms, and campaign managers in improving return on investment, optimizing advertising expenditure, and enhancing campaign performance across heterogeneous digital platforms. Furthermore, the proposed methodology establishes a scalable foundation for future AI-driven autonomous advertising systems capable of continuous learning and adaptive optimization within modern digital marketing environments. Keywords: digital advertising; campaign optimization; machine learning; predictive analytics; multi-criteria decision analysis; COPRAS; programmatic advertising; real-time bidding; decision support

Optimization of Aggregate Planning Using Linear Regression: An Integrated Approach to Demand Forecasting, Production Capacity, and Inventory Management

Rajendar Dommeti

Integrated planning plays a key role in production management, through which it is improved resource allocation, workforce utilization, inventory levels, and production capacity to meet forecasted demand while minimizing costs. This study examines the integrated planning framework for a manufacturing environment, focusing on the relationships between demand forecasting, production capacity, inventory management, and total cost optimization. Using a dataset of 300 observations, descriptive statistical analysis reveals significant variability in demand forecasts and production capacity, with mean values of 2973.9 and 3015.7 units respectively. Linear regression modeling demonstrates strong predictive performance, it has achieved an R² value of 0.9159 for the training dataset and 0.8028 for the test dataset.Correlation analysis identifies inventory level as the primary cost driver, showing a strong positive correlation of 0.81 with total cost, while demand forecast exhibits a moderate correlation of 0.41, and production capacity shows minimal direct impact at 0.24. The findings underscore the importance of balanced decision-making in aggregate planning, where effective coordination between forecasting accuracy, capacity alignment, and inventory optimization is essential for cost control and operational efficiency. This research contributes to the development of data-driven decision support systems for sustainable supply chain management. Key Words: Aggregate Planning, Linear Regression, Demand Forecasting, Inventory Optimization, Production Capacity.

Lattice-Guided Secure Edge Collaboration: An Efficient IBE-to-ABE Re- Encryption Framework over Recycled Smartphone Edge Clusters in Temporal Cloud Environments

Somasekhar Gubbala

Cloud-assisted and edge-enabled environments have become fundamental infrastructures for large-scale data storage, collaborative processing, and real-time analytics. While Ciphertext-Policy Attribute-Based Encryption (CP-ABE) provides fine-grained access control, its computational overhead renders it impractical for resource-constrained devices. Identity-Based Encryption (IBE), on the other hand, offers lightweight encryption but lacks flexible access policies. Meanwhile, the rapid obsolescence of smartphones has created a vast pool of discarded yet computationally capable devices, which can be repurposed as collaborative edge nodes. In this paper, we propose LEGSEC (Lattice-Guided Edge Secure Collaboration), a unified framework that integrates (i) IBE-to-CP-ABE proxy re-encryption for secure data sharing, (ii) lattice-theoretic resource orchestration for SLAaware task allocation, and (iii) distributed computation across recycled smartphones functioning as temporal edge clusters. Our scheme introduces a verifiable, non-interactive, and cost-aware re-encryption mechanism combined with an adaptive scheduling algorithm modeled over temporal graphs. We provide formal security analysis, algorithmic descriptions, and experimental evaluations demonstrating improved computational efficiency, reduced SLA violations, and enhanced throughput compared to baseline approaches. The proposed framework establishes a secure, sustainable, and computationally efficient paradigm for cloudassisted environments leveraging latent mobile edge resources.

Machine learning-based survival prediction in glioma using large scale registry data the importance of chemotherapy and radiation therapy management as predictive features

Zhao R, Zhuge Y, Camphausen K, Krauze A

Gliomas are the most common central nervous system tumors exhibiting poor survival, quality of life and neurological outcomes prompting significant discussion surrounding optimisation of the aggressiveness of management. The ability to estimate prognosis is crucial for both patients and providers in order to select the most appropriate treatment. Previous attempts at predicting survival outcomes have relied on clinical parameters (age, KPS, gender) and resection or methylation status and statistical models to create prognostic groups limiting survival prediction due to selection bias and tumor heterogeneity.  Machine learning (ML) allows for more sophisticated approaches to survival prediction amalgamating real world clinical, molecular and imaging data. We wanted to examine clinical parameters needed to achieve superior predictive accuracy in order to help advance guidelines for the creation and maintenance of robust large-scale glioma registries. 

The Machine learning for Predictive Maintenance in Supply Chain Management

Krishnamoorthy Selvaraj, Dr. Srinivasan Lakshmanan

It has recently come to light that one of the most important applications of machine learning in a variety of sectors, including supply chain management, is predictive maintenance. The purpose of this research is to investigate the use of machine learning strategies for predictive maintenance within the framework of supply chain management. Traditional procedures of maintenance often cause inefficiencies and interruptions in the supply chain as a result of unanticipated breakdowns of various pieces of equipment. It is possible to greatly improve both the reliability and performance of supply chain operations via the use of predictive maintenance approaches. This article starts out by giving an overview of predictive maintenance and the role that it plays in supply chain management. The issues that are presented by unanticipated equipment failures and the cascade consequences that these failures have on the supply chain are discussed. In the context of predictive maintenance, a number of different techniques to machine learning, including supervised learning, unsupervised learning, and deep learning, are analyzed and discussed. In addition to this, the study digs into data-gathering strategies, discussing topics such as sensor data, past maintenance records, and external influences that might influence the health of equipment. In addition, the article discusses the implementation issues that are associated with installing predictive maintenance systems in supply chain environments. Some of these challenges include data quality and integration, real-time decision-making, cost concerns, and others. This paper investigates the role that edge computing and industrial Internet of Things (IoT) devices play in making data gathering, analysis, and preventative maintenance more efficient.

Transforming Healthcare: The Impact and Future of Artificial Intelligence in Healthcare

Dr. Suryakiran Navath, Ph.D.

The integration of Artificial Intelligence (AI) in healthcare has ushered in a new era, transforming the industry in unprecedented ways. This manuscript delves into the profound impact of AI on healthcare systems and envisions its future trajectory. Through a comprehensive analysis of current applications, challenges, and emerging trends, this study illuminates the revolutionary changes brought about by AI technologies. From advanced diagnostics to personalized treatment plans, AI is reshaping patient care, improving operational efficiency, and fostering innovative solutions. Furthermore, the manuscript explores ethical considerations, regulatory frameworks, and the societal implications of widespread AI adoption in healthcare. By examining the intersection of technology and human well-being, this manuscript provides a holistic view of how AI is revolutionizing healthcare and offers valuable insights into the future of medicine.