Machine learning empowered computer networks

Computer Networks, Год журнала: 2023, Номер 230, С. 109807 - 109807

Опубликована: Май 3, 2023

Язык: Английский

---

Advancing UAV security with artificial intelligence: A comprehensive survey of techniques and future directions

Internet of Things, Год журнала: 2024, Номер 27, С. 101281 - 101281

Опубликована: Июль 6, 2024

Язык: Английский

---

Benchmarking Data Heterogeneity Evaluation Approaches for Personalized Federated Learning

Lecture notes in computer science, Год журнала: 2025, Номер unknown, С. 77 - 92

Опубликована: Янв. 1, 2025

Язык: Английский

---

The Impact of Pairwise and Higher-Order Complex Networks for Ai-Native Network Slicing Environment

Опубликована: Янв. 1, 2025

Язык: Английский

---

Review of Incentive Mechanisms of Differential Privacy Based Federated Learning Protocols: From the Economics and Game Theoretical Perspectives

Lecture notes in computer science, Год журнала: 2025, Номер unknown, С. 321 - 340

Опубликована: Янв. 1, 2025

Язык: Английский

---

MI-VFL: Feature discrepancy-aware distributed model interpretation for vertical federated learning

Computer Networks, Год журнала: 2025, Номер unknown, С. 111220 - 111220

Опубликована: Март 1, 2025

Язык: Английский

---

IMFL: An Incentive Mechanism for Federated Learning With Personalized Protection

IEEE Internet of Things Journal, Год журнала: 2024, Номер 11(13), С. 23862 - 23877

Опубликована: Апрель 12, 2024

Federated Learning (FL) allows clients to keep local datasets and train collaboratively by uploading model gradients, which achieves the goal of learning from fragmented sensitive data. Although FL prevents clients' being shared directly, private information may be leaked through gradients. To mitigate this problem, we combine game theory design an scheme (IMFL) based on incentive mechanism differential privacy (DP). Firstly, explore three DP variants, all are resistant deep leakage gradients (DLG) but differ in their level protection. In addition, perform convergence analysis DP. Then, with assistance theory, analyze natural state server process formulate utility function both sides under case considering attack. Finally, establish optimization problem as a Stackelberg solve for optimal strategy deriving Nash equilibrium achieve personalized Theoretical proof demonstrates that types entities can actions maximizing functions upon reaching equilibrium. Besides, extensive experiments conducted real-world demonstrate IMFL is efficient feasible.

Язык: Английский

---

A Complex Network and Evolutionary Game Theory Framework for 6G Function Placement

IEEE Open Journal of the Communications Society, Год журнала: 2024, Номер 5, С. 2926 - 2941

Опубликована: Янв. 1, 2024

Towards 6G, a key challenge lies in the placement of virtual network functions on physical resources. This becomes complex due to dynamic nature mobile environments, making design major point research. We propose framework that sees this as and collective process, presenting novel perspective which encompasses transport wireless segment aspects. The is built around an analytical modeling algorithmic tools rely systems' paradigm multiplex networks evolutionary game theory. enables capturing layered heterogeneous environment. Evolutionary theory models dynamical behavior system social where each decision influences overall outcome. Our model allows us achieve scheme optimizes 6G deployment minimizes number active computational nodes. Compared traditional centric approach, it effectively reduces interference, ensuring network's effective operation performance. Results show efficacy strategy, enabling distribution outcome dilemma, highlight potential applicability approach tackle function problem networks.

Язык: Английский

---

Federated Learning Game in IoT Edge Computing

IEEE Access, Год журнала: 2024, Номер 12, С. 93060 - 93074

Опубликована: Янв. 1, 2024

Edge Computing provides an effective solution for relieving IoT devices from the burden of handling Machine Learning (ML) tasks. Further, given limited storage capacity these devices, they can only accommodate a restricted amount data training, resulting in higher error rates ML predictions. To address this limitation, leverage and collaborate learning process through designated peer acting as device. However, transmission offloaded tasks over wireless access network poses challenges terms time energy consumption. Consequently, although collaborative diminish variance learned model, it introduces communication cost, dependent on chosen In light considerations, paper coalition formation game that proposes distributed Federated approach, where autonomously efficiently select most suitable device, aiming to minimize both their cost.

Язык: Английский

---

Unraveling trust management in cybersecurity: insights from a systematic literature review

Information Technology and Management, Год журнала: 2024, Номер unknown

Опубликована: Авг. 27, 2024

Язык: Английский

---