[CCoE Notice] Cullen College Dissertation Announcement

[Hutchinson, Inez A]

[Dissertation Defense Announcement at the Cullen College of Engineering]
Synergies of Mean-field Games in Machine Learning and 6G Communication Networks
Yuhan Kang
July 29, 2024; 9:00 AM - 10:30 AM (CST)
Location: N202 Eng Bld 1
Zoom: https://urldefense.com/v3/__https://uh-edu-cougarnet.zoom.us/j/99146061274__;!!LkSTlj0I!GksEQdWty2FcKWMpaXdXGLUhYUW837SFt3R6fsOIl-8oDd_UfzegEbn0c2Y4_3fa_fgQWh4scz2LDnyfJPARlBAiFzs$ <https://urldefense.com/v3/__https://uh-edu-cougarnet.zoom.us/j/99146061274__;!!LkSTlj0I!FCNj8gBq_29RBMwUs15szutM40K--uW4Ftg3sCDHuwPC9bATv13KvAopG_KlMPMc_Yu41dAZ0owlYk41jnZ761jX$>
Committee Chair:
Zhu Han, Ph.D.
Committee Members:
Miao Pan, Ph.D. | Hien Van Nguyen, Ph.D. | Tamer Başar, Ph.D. | H. Vincent Poor, Ph.D.
Abstract
Mean-Field Games (MFG) model decision-making in large populations where each agent's strategy has minimal impact individually but interacts with the population's average effect, simplifying the analysis compared to traditional game models. MFG uses the Hamilton-Jacobi-Bellman (HJB) and Fokker-Planck-Kolmogorov (FPK) equations to determine optimal strategies and mean field evolution, respectively. However, solving these equations becomes computationally challenging in high dimensions. This dissertation explores MFG in Machine Learning (ML) and 6G Communication Networks with three main contributions. First, we develop a Proximal-Dual-Hybrid-Gradient (PDHG) algorithm for low-dimensional MFG models. Second, we introduce a Generative Adversarial Network solution for high-dimensional MFG problems, handling up to 1,000 dimensions. Third, we apply MFG to ML and 6G networks. In vehicle-centric Mobile Crowd Sensing (MCS) networks, MFGs model cooperative and competitive behaviors, improving path planning and task selection while reducing complexity. In vehicular-based Multi-access Edge Computing (MEC) networks, MFGs enable fast data offloading, reducing End-to-End (E2E) latency. In ML, an MFG-inspired data augmentation strategy enhances test accuracy. Finally, we design incentive mechanisms for Federated Learning in satellite communications using Mean Field Evolutionary Game (MFEG) to ensure fair and efficient client participation.
[Engineered For What's Next]

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://Bug.EGR.UH.EDU/pipermail/engi-dist/attachments/20240722/c354d641/attachment.html