federated learning

A scalable and privacy-aware framework for connected vehicle systems using smartphone-based multimodal sensing

Connected Vehicle (CV) networks have emerged as key enablers of next-generation intelligent transportation systems, improving road safety and operational efficiency through real-time data exchange. However, large-scale adoption remains limited due to two fundamental challenges: 1) an unclear stakeholder value proposition and 2) persistent privacy concerns arising from continuous data collection. In parallel, conventional road condition monitoring methods are slow, expensive, and unable to provide continuous large-scale coverage. This work proposes a scalable and privacy-preserving framework that leverages smartphone-based sensing as a practical surrogate for connected vehicle data. Unlike infrastructure-dependent or single-modality approaches, the proposed method integrates widely available mobile devices with multimodal data sources to enable cost-effective, deployable large-scale road monitoring. This improves immediate real-world applicability while strengthening the value proposition needed for broader CV adoption. The research involves four main aspects. First, it has produced large-scale multimodal datasets combining smartphone inertial sensing, GPS, vision-based inputs, geographic information, and environmental conditions to capture diverse real-world driving scenarios. Second, it has used robust machine learning models to fuse heterogeneous data for accurate road anomaly detection under noisy and variable conditions. Third, it has proposed a layered privacy-preserving framework combining federated learning, contextual k-anonymity, and differential privacy to address non-IID vehicular data while ensuring strong privacy guarantees. Privacy implications are further analyzed using the IEEE Digital Privacy Model. Finally, the framework is validated through a cloud-based monitoring system and a sensing module designed to handle smartphone orientation variability in real-world deployments. In addition, the study identifies a critical gap in standardized metrics for evaluating privacy-preserving methods and proposes the need for a unified privacy score to enable systematic comparison. This research contributes a unified framework that combines scalable sensing, multimodal intelligence, and layered privacy mechanisms to balance detection performance, privacy preservation, and deployment feasibility. Overall, the proposed system establishes a scalable, privacy-aware, and deployable approach for road condition monitoring that supports intelligent transportation systems while addressing key barriers to connected vehicle adoption.

Energy-efficient federated deep learning for edge-based image recognition in resource-constrained IoT environments

Bridging Research and Practice in Federated Learning: A Multivocal Review of System-Level Architectures and Processes

Secure Federated Learning Framework with Dynamic Lightweight Cipher (Fed- DLC) for Resource-Constrained Networks and Data-Sensitive Applications

Advancements in deep learning for early breast cancer detection using artificial intelligence models in low resource settings

Breast cancer remains a leading cause of mortality among women in low- and middle-income countries (LMICs), compounded by fewer radiologists available and resources for diagnosis. A narrative review that summarizes 45 peer-reviewed publications to date from 2018 to 2025 is presented for deep learning (DL) models for mammography detection for breast cancer, targeting low-resource-critical architectures for development in LMIC settings. We compare convolutional neural networks (CNNs), hybrid CNN-support vector machine (SVM) models, recurrent/LSTM networks, and lightweight architectures such as MobileNet and EfficientNet. In quantitative synthesis the reported diagnostic accuracy for MobileNet is between 89 and 92%, supported with 128 MB RAM and 0.3 GFLOPs. Therefore it is appropriate for offline edge devices (smartphones, NVIDIA Jetson Nano). EfficientNet achieves 91–94% accuracy—though it requires stable internet to enable cloud deployment, due to its greater parameter size (5.3 million). Both architectures reduce false positives by 7 to 69% and false negatives by 9% compared to traditional computer-aided detection, though most evidence is retrospective. The most important contributions from this review are: (1) a quantitative comparison of efficiency for edge versus cloud deployment, (2) detection of dataset bias (no LMIC-specific validation exists), and (3) practical recommendations regarding infrastructure, regulatory pathways, and privacy-preserving federated learning. Limitations include cross-study heterogeneity in datasets (CBIS-DDSM, INbreast, MIAS) and evaluation protocols. We conclude that MobileNet and EfficientNet provide good trade-offs for LMICs but call for prospective trials and locally curated datasets before clinical implementation.

Diabetes Diagnosis: Comparison of Federated Learning Algorithms

FTCL: Federated Transfer Continual Learning with Outlier-Aware Adaptive EWC v2 and Episodic Replay for Personalized sEMG Gesture Recognition

Information Bottleneck with Differential Privacy for Federated Learning

Federated learning (FL) offers a promising approach to training machine learning models on decentralized data sources without direct data sharing, thus addressing privacy concerns. However, FL is still susceptible to privacy breaches and suffers from significant information loss during model aggregation, a phenomenon addressed by the information bottleneck (IB) principle. This paper proposes a novel framework that integrates the IB technique with differential privacy (DP) within the FL setting. We formulate the problem as a constrained optimization, minimizing information loss while simultaneously satisfying DP guarantees. Our approach utilizes a compressed representation of local data, learned through an IB objective, and introduces noise to protect individual data points, ensuring privacy. The core contribution lies in the synergistic combination of these two techniques, leading to enhanced privacy protection and improved model accuracy compared to standard FL. We demonstrate the effectiveness of our framework through a theoretical analysis and outline potential implementation strategies. The primary goal is to achieve a balance between model performance and privacy preservation, a critical aspect often overlooked in current FL methodologies. The theoretical framework provides a foundation for future research and practical deployment in privacy-sensitive applications.

Decentralized Federated Learning with Byzantine Fault Tolerance via Consensus-Based Proofs

Federated learning (FL) offers a promising approach to training machine learning models on decentralized data sources without directly exchanging the data itself. However, existing FL systems are susceptible to Byzantine attacks, where malicious participants can inject faulty model updates, compromising the global model's integrity. This paper proposes a novel decentralized federated learning framework incorporating Byzantine Fault Tolerant (BFT) consensus protocols. Our system utilizes cryptographic consensus mechanisms to validate and authenticate model updates from each participant before aggregation, thereby mitigating the risks posed by Byzantine nodes. The core innovation lies in the integration of FL with robust BFT consensus, ensuring secure and reliable model training even when faced with adversarial behavior. We introduce a framework utilizing verifiable computation and consensus-based proofs to achieve this. This approach allows for the detection and rejection of malicious updates, ultimately leading to a more trustworthy and resilient global model. The system is designed for scalability and adaptability, addressing key challenges in practical FL deployments. The key mathematical concepts underlying the system are represented through the following notation: (x_i, m_i), where x_i represents the data sample from participant i, and m_i represents the model update generated by participant i. The aggregation function is denoted as (Σ_{i=1}^K (α_i * m_i)), where α_i represents the learning rate for participant i, and K is the total number of participants. The BFT consensus protocol relies on a threshold number of participants (T) to reach agreement, and the proof of correctness is represented as P(m_true, m_agg), where m_true is the true global model, and m_agg is the aggregated model. Byzantine faults are represented as f_i, where f_i is the faulty model update from participant i. The probability of a successful consensus is denoted as P_success. The security level is characterized by the parameter β, representing the probability of successfully detecting a Byzantine update.

Distributed Federated Learning with Differential Privacy and Byzantine Fault Tolerance

Federated Learning (FL) offers a promising paradigm for training machine learning models on decentralized data sources without directly exchanging data. However, existing FL frameworks are susceptible to various vulnerabilities, including privacy breaches through information leakage and attacks from Byzantine clients attempting to compromise the learning process. This paper proposes a novel distributed FL framework that integrates differential privacy (DP) and Byzantine fault tolerance (BFT) mechanisms to address these challenges comprehensively. Our approach employs advanced DP techniques to rigorously limit the information revealed by individual clients during model updates, while simultaneously utilizing BFT algorithms to detect and mitigate the influence of malicious or faulty clients. The resulting system demonstrates improved security, enhanced privacy guarantees, and robustness against adversarial attacks, making it a significant advancement in the field of secure and reliable distributed learning. The core of our work lies in the synergistic combination of these two crucial techniques, providing a layered defense against potential threats in FL environments. We detail the mathematical formulations underlying our approach and provide a theoretical analysis of its performance.

Federated discretization for supervised classification

Abstract Several classifiers operate exclusively on discrete variables, requiring continuous attributes to be preprocessed through discretization. Supervised discretization addresses this need by partitioning the domain of continuous variables into intervals that define a discretization policy. Federated learning, in turn, enables the collaborative training of global models without sharing raw data: users train local models and communicate their parameters to a central server, which aggregates them into a global model. A key challenge in this setting is achieving performance comparable to centralized methods while minimizing communication overhead. In this work, we propose a federated discretization framework that computes a global discretization policy in a single communication round. Each user estimates Gaussian mixture models for the continuous variables and transmits their parameters-referred to as representatives-to the server. The server then applies a discretization algorithm to these representatives to derive a global policy, which is subsequently distributed back to the users. The proposed framework accommodates multiple discretization strategies; specifically, we present federated adaptations of equal-width, equal-frequency, entropy-based, and kernel density-based methods. Experiments conducted on ten benchmark datasets demonstrate that the proposed framework achieves performance comparable to centralized discretization approaches, with the kernel density-based and equal-frequency variants yielding the most competitive results.

Distributed Graph Learning via Federated Bayesian Networks

This paper presents a novel approach to distributed graph learning utilizing Federated Bayesian Networks (FBNs). The core challenge in training large graph neural networks (GNNs) lies in the substantial computational resources required, often necessitating centralized training environments. Federated Bayesian Networks offer a decentralized solution, enabling learning across multiple clients without direct data sharing. The proposed method involves local training of Bayesian Networks on individual client graph subsets, followed by parameter aggregation by a central server to refine a global Bayesian Network model. This architecture addresses the limitations of traditional GNN training while prioritizing data privacy and mitigating computational demands. The key innovation lies in the synergistic combination of federated learning principles with the probabilistic inference capabilities of Bayesian Networks, resulting in a robust and scalable framework for distributed graph learning. This approach demonstrates the potential for efficient learning from decentralized graph data sources.