• 8:25 am - Welcome & Overview of Workshop 

• 8:30 am - Invited Talk 1 - Jia (Kevin) Liu, Ohio State University

Federated Multi-Objective Learning

Abstract: In the first part of this talk, we will introduce a new federated learning paradigm called federated multi-objective learning. In recent years, multi-objective optimization (MOO) emerges as a foundational problem underpinning many multi-agent multi-task learning applications. However, existing algorithms in MOO literature remain limited to centralized learning settings, which do not satisfy the distributed nature and data privacy needs of such multi-agent multi-task learning applications. This motivates us to propose a new federated multi-objective learning (FMOL) framework with multiple clients distributively and collaboratively solving an MOO problem while keeping their training data private. Notably, our FMOL framework allows a different set of objective functions across different clients to support a wide range of applications, which advances and generalizes the MOO formulation to the federated learning paradigm for the first time. For this FMOL framework, we propose two new federated multi-objective optimization (FMOO) algorithms called federated multi-gradient descent averaging (FMGDA) and federated stochastic multi-gradient descent averaging (FSMGDA). Both algorithms allow local updates to significantly reduce communication costs, while achieving the same convergence rates as those of the their algorithmic counterparts in the single-objective federated learning. Our extensive experiments also corroborate the efficacy of our proposed FMOO algorithms.

• 9:00 am - Invited Talk 2 - Cesar Uribe, Rice University 

Sparse factorization of the square all-ones matrix and Consensus problems for next-generation networks

Abstract: In this presentation, we study sparse factorization of the (scaled) square all-ones matrix of arbitrary order. We study sequences of graphs satisfying the finite-time consensus property (i.e., iterating through such a finite sequence is equivalent to performing global or exact averaging). We introduce the concept of hierarchically banded matrices and propose two types of hierarchically banded factorization. Finally, we discuss the application of the proposed sparse factorizations to the decentralized average consensus problem and decentralized optimization for next-generation communication networks. 

• 9:30 am - Invited Talk 3 - Aggelos Blestas, Rutgers University

ML-based Localization and Asynchronous Message Passing Inference for Batteryless IoT Networks

Abstract: Enabling machine learning (ML) and inference algorithms in next Generation networks necessitates rethinking several established assumptions on networks, resources, and infrastructures, such as availability of location information and time-synchronized, coordinated operation for (synchronous) inference. Especially for the coming, completely batteryless Internet of Things, where sensor terminals are resource-constrained, facilitating coordinated, synchronous inference is not scalable, while estimating location information is challenging. The first part of the talk puts forth a deep feed forward neural network (NN) that performs regression for tag localization, using phase-based measurements as input from a custom multistatic RFID interrogation setup. Experimental results in real-world scenarios offer mean absolute error of a few centimeters, even though the NN was trained with synthetic (i.e., simulation) data. Given that ML-based techniques are limited by the quantity and quality of the data, this work offers a practical training methodology for ML-enabled wireless networks. The second part of the talk offers design and implementation of in-network inference, using message passing among ambiently powered wireless sensor network (WSN) terminals. The stochastic nature of ambient energy harvesting dictates intermittent operation of each WSN terminal and as such, the message passing inference algorithms should be robust to asynchronous operation. Examples from Gaussian belief propagation and average consensus (AC) are provided, along with the derivation of a statistical convergence metric for the latter case. Interestingly, it is shown that there are divergent instances of the in-network message passing algorithms that become convergent, under asynchronous operation. 

• 10:00 am - Coffee Break

• 10:30 am - Invited Talk 4 - Jun Zhao, Nanyang Technological University

AI for Communication Networks

Abstract: In this talk, I will share AI-driven innovations for communication networks, with a focus on semantic communication (SemCom), reinforcement learning (RL), federated learning (FL), and secure fine-tuning methods. SemCom systems are enhanced through post-deployment fine-tuning, enabling greater adaptability and efficiency in data transmission across diverse environments. Reinforcement learning plays a key role in optimizing resource allocation within mobile edge computing (MEC), particularly in real-time video transmission, and in improving decision-making in multi-agent systems. In federated learning, AI techniques are employed to address communication and computation challenges, balancing bandwidth and energy constraints in wireless networks. Additionally, secure offsite-tuning techniques are introduced to address the computational and privacy concerns of fine-tuning large models on edge devices. By combining offsite tuning with physical-layer security, we optimize resource allocation and compression ratios while preserving privacy and minimizing energy costs. These AI techniques significantly advance communication network performance, offering improvements in resource efficiency, flexibility, and security.  

• 11:00 am - Invited Talk 5 - Marios Kountouris, EURECOM

• 11:30 am - Invited Talk 6 - Matthew Dwyer, United States Army Research Laboratory

• 12:00pm - Closing Workshop Remarks & Workshop Adjourns