federated learning

Distributed Learning with Federated Bayesian Networks

This paper presents a novel approach to distributed learning leveraging Federated Bayesian Networks (FBNs). Traditional federated learning methods struggle with capturing complex, non-independent data distributions, often leading to suboptimal global models. We introduce a framework where multiple devices collaboratively learn a Bayesian Network model through decentralized updates, explicitly addressing these dependency challenges. The core idea is to allow devices to share probabilistic models and inference results, creating a system that adapts to local data characteristics while maintaining data privacy. The framework incorporates techniques for managing network topology, handling heterogeneous data distributions, and ensuring convergence of the collaborative learning process. We demonstrate the potential of this approach through a theoretical analysis and outline key considerations for practical implementation, highlighting its advantages over conventional federated learning systems. The primary contribution lies in the systematic application of Bayesian Networks to federated learning, providing a robust mechanism for learning from correlated data sources.

AI for EV battery health: Statistical meta-analysis and Autonomous Intelligence Pyramid

The transportation sector accounts for approximately 21% of global CO 2 emissions ( ≈ 8.4 Gt CO 2 in 2024), and the electrification of road transport is a critical lever for decarbonisation. Annual global EV battery deployment reached 1.2 TWh in 2025, making accurate battery health monitoring essential for both performance and sustainability. This review, conducted via a PRISMA-based protocol over 531 peer-reviewed studies (2018–2026), provides a statistical meta-analysis of all major AI families deployed in battery management: classical machine learning, ensemble methods, deep sequential networks, hybrid CNN architectures, Transformers, Physics-Informed ML (PIML), Explainable AI (XAI), Federated Learning, and Large Battery Foundation Models. A DerSimonian–Laird random-effects meta-analysis with 95% confidence intervals and I 2 heterogeneity statistics confirms that CNN–LSTM achieves a pooled RMSE of 0.47% (95% CI: 0.41%–0.53%, d = 3.42 vs EKF baseline) and Transformer architectures reach 0.42% (0.36%–0.48%, d = 3.78 ). The Foundation Model group, based on only nine studies at TRL 2–3, yields a preliminary pooled RMSE of 0.42% that is sensitivity-dependent on a single influential study; this figure should not be directly compared with the more robustly supported Transformer estimate. Critical limitations of all emerging methods are explicitly characterised. Environmental and social impacts of battery energy systems, including CO 2 lifecycle analysis, are discussed. An author-proposed five-layer Autonomous Battery Intelligence Pyramid is presented as a speculative research roadmap, with a dedicated implementation pathway discussion, clearly distinguished throughout from experimentally validated findings.

Cross-Chain Federated Learning Framework over Blockchain for Secure and Interoperable Decentralized AI

In recent years, the rapid growth of distributed artificial intelligence (AI) and blockchain technology has led to new opportunities for building secure, transparent, and privacy-preserving learning systems. Federated Learning (FL) enables multiple users or organizations to collaboratively train a global AI model without sharing their private data, while Blockchain provides immutability, traceability, and decentralized trust. However, most existing blockchain-based FL systems are limited to a single network, lacking interoperability and scalability across multiple chains. This review paper explores the emerging concept of Cross-Chain Federated Learning (CCFL), which integrates federated learning with cross-chain blockchain communication to achieve secure and interoperable decentralized AI. The paper discusses existing research works, current architectures, algorithms, and cross-chain mechanisms, identifying key challenges such as model verification, communication overhead, and data integrity. Furthermore, it highlights how the proposed framework addresses these challenges by using smart contracts, cryptographic hashing, and relayer-based synchronization.The study concludes that integrating FL with cross-chain blockchain technology can significantly enhance privacy, security, and collaboration among diverse AI systems, paving the way for next-generation decentralized intelligence. Keywords— Federated Learning, Blockchain, Cross-Chain Communication, Decentralized AI, Data Privacy, Smart Contracts, Secure Aggregation, Interoperability

Few-Shot Federated Learning for State-of-Charge Prediction Across Privacy-Isolated Personal Mobile Devices

Accurate state-of-charge (SOC) prediction on personal mobile devices requires a model that can represent heterogeneous discharge mechanisms, transfer knowledge across users, and adapt from scarce target records while raw telemetry remains privacy isolated. This study proposes a module-decomposition-driven SOC prediction model that separates baseline drain, synchronous state response, and asynchronous multi-scale event memory within one discharge-dynamics formulation. The same three semantic coefficient blocks remain aligned through local source learning and round-wise sample-count-weighted federated aggregation. Target-side personalization then adjusts one locally derived personal discharge-rate scalar while preserving the shared temporal response, and forward SOC propagation produces threshold-based time-to-empty (TTE) estimates. A user-held-out evaluation on a public small-sample smartphone-log dataset uses eight target partitions and 447 final evaluation origins. Personalized fine-tuning reduced TTE MAE by 67.07% and RMSE by 71.88%, yielding an R2 of 0.936. The resulting task-specific architecture connects module-structured data processing, federated synthesis, few-shot adaptation, and local TTE inference while raw usage logs remain within user-specific data boundaries.

Federated Learning with Differential Privacy: A Distributed Approach

This paper presents a novel approach to distributed machine learning, termed Federated Learning with Differential Privacy (FLDP), which addresses the critical challenge of privacy preservation in collaborative model training. Traditional federated learning techniques, while mitigating the risks associated with centralized data collection, still expose model updates, potentially revealing sensitive information about the underlying data. FLDP overcomes this limitation by directly applying differential privacy to the model aggregation process. Specifically, we introduce a locally sensitive differential privacy mechanism that injects noise into each node's model update before aggregation. This ensures that the impact of any single node's contribution on the global model is bounded, effectively protecting individual data privacy. The proposed method allows for reliable model convergence while rigorously upholding privacy guarantees. We demonstrate the feasibility and effectiveness of FLDP through a theoretical analysis and outline a practical implementation strategy. The key contribution of this work lies in the shift from applying differential privacy solely to data training to directly incorporating it into the model aggregation phase, creating a more robust and privacy-preserving distributed learning system.

Systematic Review on AI-Powered UAVs: The Role of Artificial Intelligence in UAV Evolution and Applications Expansion

Unmanned aerial vehicles (UAVs) are progressively evolving from remotely operated platforms into intelligent autonomous systems. This research addresses the role of Artificial Intelligence (AI) in advancing UAV capabilities and expanding their applications across diverse sectors. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 methodology, studies retrieved from Scopus and relevant academic books and book chapters were screened, resulting in 217 publications retained for final analysis. The analysis introduces a three-layer framework linking AI techniques, UAV functional capabilities, and application domains. The first layer covers the list of adopted AI and ML approaches in the UAV applications. The second layer maps these approaches to key UAV capabilities, including perception, autonomous navigation, control and stability, swarm coordination, communication, and energy optimization. The third layer examines applications in agriculture, logistics, disaster response, environmental monitoring, surveillance, defense, and wireless network systems. The findings show that deep learning enhances aerial perception, reinforcement learning supports adaptive navigation and control, federated learning improves distributed intelligence, and swarm intelligence enables cooperative multi-UAV missions. Despite these advances, AI-enabled UAVs still face challenges related to energy consumption, onboard computation, data availability, communication reliability, safety, ethics, privacy, and regulation. Future progress is expected to be driven by edge AI, Tiny Machine Learning (TinyML), quantum-inspired optimization, explainable artificial intelligence (XAI), human-AI collaboration, and robust swarm coordination. Overall, this review provides a structured synthesis of AI-enabled UAV research and identifies key directions for future innovation.

Deep learning for sustainable development: cross-disciplinary applications and responsible AI

Global climate change, energy shortages, food security crises, and urban sprawl pose multifaceted governance challenges for sustainable development. Leveraging its capacity for hierarchical representation learning from high-dimensional, heterogeneous data, deep learning has emerged as a core digital technology enabler for advancing the Sustainable Development Goals (SDGs). Employing a systematic literature review methodology and drawing on open-access publications from Google Scholar (2018-2026), this paper examines deep learning applications across four key domains: climate action, sustainable energy, smart agriculture, and smart cities. Through four comparative tables analyzing model suitability, data types, application outcomes, and challenges, the study identifies key bottlenecks in technology deployment from a "Responsible AI" perspective. It identifies three major industry trends: the large-scale deployment of physics-AI hybrid modeling, the growing dominance of deep reinforcement learning in energy dispatch, and the integration of federated learning with edge AI for distributed ecological monitoring. Current implementation efforts remain constrained by four critical issues: regional data divides, a lack of model interpretability, algorithmic bias, and the high carbon footprint associated with large-scale models. By outlining corresponding optimization pathways, this paper provides a theoretical framework and practical guidance for green AI and sustainable digital governance.

A task-oriented review of deep learning for medical image segmentation from 2015 to 2026

Segmentation of medical images is a fundamental step towards quantification and clinical decision support. Many approaches have been proposed and successfully demonstrated on benchmark datasets. However, reliable application of current approaches in real clinical practice remains challenging. Medical image segmentation is not a single task but rather a group of different tasks with different input and output characteristics, which call for dedicated solutions. Task difficulty and failure modes vary with anatomical priors, boundary ambiguity, and clinically meaningful error types. Therefore, rather than providing only a method summary, this review presents a task overview of medical image segmentation research from 2015 to 2026. We organize the literature according to clinical task properties and examine how different methodological choices influence benchmark results and real-world applicability. We categorize existing work into major task families, including organ segmentation, lesion and tumor segmentation, vascular segmentation, histopathology and cellular segmentation, and multi-task and cross-domain settings. For each category, we summarize representative modeling strategies and identify common challenges specific to each task. We further review widely used public datasets and evaluation metrics. In particular, we analyze how overlap, boundary, topology, and instance-level metrics capture different clinical priorities. Finally, we discuss challenges such as annotation burden, inter-institution variability, and clinical reliability. We then discuss emerging research directions, including foundation models, multi-modal and multi-task learning, and privacy-preserving federated collaboration, which may contribute to improved public health and clinical decision-making.

Emerging Trends in Artificial Intelligence: A Review of Transformative Technologies and Applications

The current trends in Artificial Intelligence technologies continually transform multiple business sectors while changing how humans connect with machines. Contemporary technology views AI as its essential component instead of remaining as a separate specialized study field. A review of primary AI trends coupled with applications and obstacles in modern AI research served as the focus of this academic paper. Healthcare decision transparency integrates with financial needs to enable smart infrastructure deployment by connecting Explainable AI (XAI) with Federated Learning under Edge Computing framework. The barriers to progress include major issues with scalability together with interoperability standards and ethical needs for organizational governance. Interdisciplinary teamwork constitutes an absolute requirement to construct responsible AI systems together with dependable deployment frameworks as demonstrated through the term paper. The review combines recent research findings with guidance on new investigation tracks for developing all-encompassing AI systems which both spread far-reaching outcomes and follow sustainability requirements.

A Comprehensive Analysis of Dynamic Task Scheduling for Cloud Computing Environment

Cloud computing has revolutionized the way computational resources are made available through scalable, flexible, and on-demand services. As promising as it is, efficient task scheduling remains an important challenge because of varied workloads, heterogeneous infrastructures, and multi-objective optimization requirements for cost, execution time, and power efficiency. Legacy scheduling approaches frequently fail to meet these demands, and thus the idea of hybrid frameworks fusing artificial intelligence (AI) and optimization techniques has been developed. In this paper, dynamic task scheduling techniques that use deep learning, metaheuristic optimization, and heuristic algorithms to be more efficient are discussed. Specific focus has been placed on energy-efficient models like adaptive Particle Swarm Optimization (PSO) and multi-objective scheduling models with a balance between performance and sustainability. The developed AI-based model utilizes deep learning to predict the workload, optimization techniques to achieve multi-objective trade-offs, and reinforcement learning to adapt in real-time. The research helps in the development of multi-objective, adaptive, and sustainable task scheduling for cloud platforms with future prospects on federated learning, edge/fog integration, and security-conscious scheduling techniques.

Decentralized Federated Learning with Differential Privacy using Threshold Cryptography

Federated learning (FL) presents a promising approach to training machine learning models on decentralized data sources while preserving user privacy. However, traditional FL methods often rely on a central server, introducing a single point of failure and potential privacy risks. This paper proposes a novel decentralized federated learning system that leverages threshold cryptography and differential privacy to address these concerns. The core claim is that protecting user privacy in federated learning remains a significant challenge, and this system provides a robust solution. The system operates through a series of distributed rounds where participants collaboratively update model parameters using threshold cryptography to ensure secure aggregation and differential privacy to mitigate individual data exposure. This approach eliminates the need for a central server, enhancing both privacy and security. The system is designed for scalability and adaptability, making it suitable for various decentralized data scenarios. This work contributes to the development of more secure and privacy-preserving federated learning solutions.

Defending Poisoning Attacks in Federated Learning Under System and Data Heterogeneity