International Journal of Intelligent Systems and Data Science

Vol. 1 No. 2

2026
June 2026 · 14 articles · Volume 1, Issue 2
Back to archives
ISSUE OVERVIEW

Intelligent systems. Real-world impact.

This issue presents research advancing intelligent computing, data science, cloud infrastructure, cybersecurity, and autonomous systems. The featured articles explore machine learning optimization, scalable Bayesian methods, DevOps automation, virtualized infrastructure evaluation, enterprise SaaS architectures, and AI-driven anomaly detection. Additional contributions address collaborative robotics, automated analytical workflow generation, computational performance optimization, cloud access-control verification, container security, healthcare interoperability security, and scientific mobility analysis. Collectively, these studies highlight innovative approaches for building secure, scalable, and intelligent systems that support modern data-driven technologies and decision-making.

Published

June 2026

Contributions

14 articles

Access

Open access

June 2026 · 14 articles
Issue Cover
EXPLORE THIS ISSUE
Table of contents
14 articles

Explore the research published in this issue of ijisds.

IN THIS ISSUE
eBPF‑Secure: A Lightweight eBPF‑Based Confinement and Observability Framework for Container Security

Shiva Bommakanti

As container adoption accelerates, the need for runtime security and transparent observability has never been greater. Traditional Linux confinement primitives, namespaces, cgroups, seccomp, and LSMs are powerful but often complex to combine and difficult to extend. In this work, we introduce extended Berkeley Packet Filter (eBPF) ‑ Secure, a unified eBPF‑based framework that dynamically injects security and observability logic into the kernel without requiring recompilation or reboot. By attaching eBPF programs to key syscall and LSM hooks, eBPF‑Secure enforces container-specific policies for filesystem, network, and IPC interactions, while simultaneously gathering rich audit data into kernel‑space maps for real‑time monitoring. Two prototypes, an application sandbox and its container‑focused successor, demonstrate that eBPF‑Secure can model least‑privilege confinement, bridge the semantic gap between policy and enforcement, and deliver precise security controls with only modest performance overhead. Our evaluation against established benchmarks and an informal attack analysis shows eBPF‑Secure’s ability to harden container deployments while providing operators with fine‑grained visibility into runtime behavior, paving the way for more secure, observable, and extensible container platforms.

PDF
The Impact of Wing Weight Ratio on Rotocopter Flight Time: A Systematic Study

Aakash Gadh

This study systematically investigates the effect of wing weight distribution on the descent time of paper rotocopters, a model system for passive rotary flight. Using a controlled drop height of 3.0 m and a fixed total added mass of 2.0 g (four 0.5 g paper clips), six configurations of left-to-right wing weight ratio (100:0, 75:25, 60:40, 50:50, 25:75, 0:100) were tested. Each configuration was replicated three times, with descent time measured manually from release to ground contact. The symmetric configuration (50:50) yielded the longest mean flight time (2.38 s), while extreme asymmetries (100:0 and 0:100) reduced flight time to approximately 1.85–1.89 s. A quadratic regression model (R2 = 0.94) shows an optimum near 55% weight on the left wing, suggesting that slight asymmetry may be tolerated without severe performance loss. These results support the theoretical expectation that balanced torque and moment of inertia maximise aerodynamic stability and descent duration. However, manual timing, release inconsistency, and environmental airflow introduced variability, highlighting the need for refined measurement methods. The findings have implications for biomimetic micro air vehicles, deployable sensors, and low-cost passive descent systems, demonstrating that even simple paper models can yield quantitative insights into rotational stability and drag interactions.

PDF
Formal Verification of Cloud RBAC Policies for Secure Google IAM Configurations

Jitendra Gupta

Misconfigured access control policies remain a leading cause of security breaches in cloud environments, yet administrators lack systematic tools for rigorous validation of complex policy sets. This paper introduces a formal verification framework for Google Cloud Identity and Access Management (IAM) that precisely models its role-based access control (RBAC) semantics, including hierarchical resource structures (organization, folders, projects, resources), policy inheritance rules, and conditional bindings. The proposed RBAC\textsubscript{GCP} model is operationalized as a transition system, enabling automated model checking against security properties expressed in Computation Tree Logic (CTL). We implement the framework using the NuSMV symbolic model checker and evaluate it on three realistic Google Cloud scenarios: Cloud Pub/Sub messaging, Cloud Storage, and Compute Engine virtual machines. The case studies demonstrate the framework's ability to detect unintended permission inheritance, cross-branch privilege leaks, violations of least privilege, and separation-of-duty constraints. When a property fails, the model checker generates a concrete counterexample that pinpoints the exact member, resource, permission, and inheritance path responsible, transforming verification into an actionable remediation tool. The results show that formal verification can serve as an effective cybersecurity tool for proactive assessment of cloud access configurations, supporting compliance audits and preventing data breaches before policy deployment.

PDF
Differential Protocol Parsing Exploits in Distributed Health Information Exchange Networks: Language-Theoretic Security Analysis of Federated EHR Interoperability Infrastructure

Jitendra Gupta

Distributed health information exchange (HIE) networks enable patient data portability across heterogeneous providers via standards like FHIR. However, independent implementations introduce semantic parser divergences that adversaries can exploit to violate data integrity or disclose protected information. This paper presents a differential fuzzing methodology to systematically identify parser misinterpretations across distributed FHIR server implementations a critical component of cloud-based healthcare data exchange. We introduce FHIR Garden, a containerized testbench that compares how seven open-source and commercial FHIR implementations interpret identical patient data structures. Empirical analysis uncovered 59 parser differentials, including incompatible JSON/XML parsing, numeric precision handling, and Unicode normalization discrepancies. Exploit chains were constructed to achieve selective data transmission through specific server combinations while causing rejection in others. Vulnerabilities in OpenEMR's CCDA processing led to timestamp stripping, vaccine description mutation, and catastrophic decimal truncation that can impair clinical decision support. Infrastructure reconnaissance further identified 1,089 publicly accessible FHIR servers, with minimal enforcement of SMART-on-FHIR authentication, exposing potential unauthorized access to patient datasets. Our findings demonstrate that interoperability standardization inadvertently expands the attack surface in federated cloud healthcare architectures, necessitating security-aware implementation specifications and dynamic verification frameworks to maintain data fidelity across distributed health information networks.

PDF
Scalable Bayesian Learning with Localized Global Approximations

Dhanush Gopal Battina

This paper addresses the challenge of applying Bayesian parameter learning to extensive datasets and complex models, a scenario where traditional Expectation Propagation (EP) faces significant memory limitations. We introduce Stochastic Expectation Propagation (SEP), an innovative algorithm designed to maintain a global posterior approximation while leveraging local update mechanisms akin to EP. This method effectively combines the accuracy characteristics of EP with a substantial reduction in memory footprint, scaling inversely with dataset size. Evaluations across various datasets demonstrate that SEP achieves performance comparable to full EP, providing an efficient pathway for robust Bayesian inference in large-scale data environments.

PDF
VMPlaceS Enables Scalable Evaluation of Virtual Machine Placement Strategies Using a High-Fidelity Simulation Framework

Apeksha Bhuekar

Virtual machine (VM) placement plays a critical role in improving resource utilisation, reducing operational costs, and maintaining Service Level Agreement (SLA) compliance in cloud computing environments. However, evaluating VM placement algorithms in real infrastructures is often expensive, time-consuming, and difficult to reproduce at scale. This paper presents VMPlaceS (VMPS), a high-fidelity simulation framework built on SimGrid for the development, testing, and comparative analysis of dynamic VM placement strategies. VMPS provides a configurable environment for modelling large-scale cloud infrastructures, dynamic workloads, VM migrations, and resource contention while supporting reproducible experimentation through synthetic workload generation. The framework incorporates an event-driven architecture consisting of initialisation, workload injection, and trace analysis phases, enabling detailed monitoring of system behaviour and algorithm performance. To demonstrate its capabilities, three representative placement approaches are implemented and evaluated: the centralised Entropy algorithm, the hierarchical Snooze framework, and the distributed DVMS strategy. Experimental results show that VMPS closely reproduces real-world behaviour, with simulation outcomes differing from in-vivo executions by a median of approximately 12%. Scalability studies further indicate that distributed placement approaches achieve superior responsiveness and lower violation times in large-scale environments. The proposed framework provides researchers and practitioners with a practical and extensible platform for evaluating VM placement solutions under realistic cloud conditions.

PDF
Evaluating Data Balancing Techniques and Feature Selection for Improving Classification Accuracy in URL Access

Santosh Vemula

This study examines the impact of data balancing techniques and feature selection on the classification accuracy of URL access requests in corporate environments. Undersampling and oversampling were evaluated using J48, Random Forest, and REP Tree classifiers, with oversampling achieving better performance by preserving important data patterns. Random Forest and REP Tree showed the highest classification accuracy, while redundant patterns were removed to streamline the dataset without significantly affecting performance. The study also applied Rough Set Theory (RST) for feature selection, reducing attributes from 12 to 9, which decreased computational time while maintaining or slightly improving accuracy. Robustness testing, conducted by excluding URL and IP address features, reduced accuracy but still produced practical rules for identifying malicious or unauthorized access requests. The proposed framework supports dynamic, context-aware URL access classification beyond static whitelist/blacklist methods, improving efficiency and reliability in enterprise security systems. Future work will focus on real-world validation and advanced classification and feature extraction techniques.

PDF
Mapping Scientific Frontiers: Network Embeddings Reveal Hidden Structures in Global Research Mobility

Akhil Veluru

The linkage of research organisations going far beyond geography holds significance in understanding global scientific mobility and in developing a better communication mechanism; the need for an analytical framework is felt. This study introduces a representation learning method in which researchers’ co-affiliation trajectories are treated as sentence-like sequences. The authors employ a skip-gram with negative sampling (SGNS) model that embeds institutions into a high-dimensional space. In contrast to traditional approaches that use geographical distance, our learned embeddings capture the multi-dimensional cognitive proximity, organisational proximity, cultural proximity and linguistic proximity we find in the actual movements of millions of researchers. Evidence suggests that the cosine similarity between institutional embeddings can explain more than two times the variance in observed researcher flows as compared to distance. When incorporated into an augmented gravity model, the embedding-based predictor significantly outperforms in predictive accuracy for both intra-national and international mobility, achieving Pearson correlations of 0.79 and 0.76, respectively, vs 0.54 and 0.49 for geographic distance. The UMAP mapping visualisation helps to interpret the clusters corresponding to language groups, historical legacies, local academic ecosystems, eg, co-location of French-speaking institutions across continents or co-location at the level of states in the US. Network embeddings can successfully recover the hidden structure underlying international scientific mobility. Network embeddings, therefore, in development, provide a new and powerful data-driven tool for science policy and research on knowledge diffusion. 

PDF
Semantic Synthesis of Analytical Pipelines: Automated Workflow Composition for Predictive Data Science

Muneebuddin Syed

The proliferation of data science libraries has created a complexity barrier for practitioners seeking to construct effective analytical workflows. This paper introduces a constraint-driven framework for automated pipeline generation that operates at a semantic abstraction layer above technical implementations. By modeling domain-specific ontologies encompassing data manipulation, feature engineering, and machine learning operations, the system synthesizes executable Jupyter notebook workflows through SAT-based reasoning. An alternative Answer Set Programming backend extends synthesis capabilities for knowledge-intensive search problems. Evaluation across three canonical datasets housing price regression, survival classification, and sentiment analysis demonstrates the framework's ability to generate valid workflows while revealing trade-offs between automation flexibility and user control. The approach enables domain experts to rapidly prototype analytical solutions without deep programming expertise, accelerating the data science lifecycle from hypothesis formulation to model evaluation.

PDF
Architecture-Aware Synthesis of Fused Linear Algebra Kernels

Shlok Shah

High-performance code generation from high-level array-oriented prototypes remains challenging due to the tight coupling between memory hierarchy behavior, loop organization, and platform compilers in modern architectures. This paper presents a compiler autotuner pipeline that lowers numerical kernels from a high-level prototype to C, applies loop restructuring for parallelism and locality, and then performs empirical search over transformations such as fusion, unroll-and-jam, vectorization, and alignment to select architecture-optimal variants. Central to the approach is composing multiple dense operations into a single fused kernel to minimize data movement and improve cache residency compared to sequential library calls. An annotation-driven tuning workflow systematically explores implementation choices and search heuristics to realize portable performance across target systems without manual rewrite cycles. Evaluations on representative kernels (e.g., VADD, ATAX, GEMVER, GESUMMV, BiCG) show consistent speedups over baseline C and vendor-tuned libraries, while also highlighting cases where library baselines remain competitive, motivating hybrid generation strategies. The results highlight a systems-centric co-design of compiler analysis and empirical tuning. The abstract should clearly state the research problem, objectives, theoretical grounding, methodology, key findings, and principal managerial implications. Emphasis should be placed on the relevance of the study to adaptive management practices and business intelligence–driven decision-making. Avoid citations, undefined acronyms, and excessive technical detail.

PDF
Showing 1 to 10 of 14 articles
Research, freely accessible

Read the articles online or download individual PDFs from this issue.