Room: MR308-309
Abstract
In recent years, the advent and ongoing refinement of Generative Artificial Intelligence (GenAI) technologies, including large-scale models, generative adversarial networks, and denoising models, has significantly expanded the potential applications of this transformative technology. The GenAI agent is capable of exhibiting human-like responses in question-answering tasks, with a singular focus on the domain of computer science. As AI models grow in complexity and size, traditional cloud-based approaches often struggle with issues like latency, bandwidth constraints, and data privacy concerns. Edge GenAI seeks to address these challenges by enabling GenAI models to run locally on edge devices, such as smartphones, IoT sensors, or embedded systems, closer to where the data is generated. Edge devices can support distributed model deployment and flexibly share computing loads with other devices or central servers to collaboratively complete generation tasks. The advent of hosting generative AI on edge devices introduces a range of novel problems that are ripe for exploration, combining the challenges of inter-model communication, distributed computing, distributed training of AI models, etc. Currently, the development of network technology to support mobile edge GenAI is still in its infancy. Efficient communication technologies, such as encoding technologies, transmission protocols, and joint resource allocation issues between distributed GenAI models still urgently require insightful academic contributions. Topics of interest include but are not limited to
- Distributed/federated training of GenAI
- Scalable Architectures of Edge Models
- Information and Communication Technologies of GenAI
- GenAI for Network Scheduling and Resource Allocation
- Model Compression and Knowledge Distillation of GenAI
- GenAI in Intelligent Transportation
- GenAI enabled Metaverse and Digital Twins
- GenAI for Healthcare
- Privacy and Security challenges in GenAI
- Edge GenAI models’ Update
- Multi-modal GenAI
- Fine-tuning Technologies of GenAI Models
Organizers
- Abdellah Chehri, Royal Military College of Canada, Canada
- Mona Jaber, Queen Mary University of London, UK
- Ruikang Zhong, Queen Mary University of London, UK
WS01-S1:
Time: 14:00 – 15:30
Room: MR308-309
Chair: Abdellah Chehri
Presentations:
1571233644: Quantum-Neuromorphic Hybrid Architecture for Energy-Efficient Multi-Agent Coordination in Distributed Healthcare Networks
Muthukumarapandian Chandrasekaran
1571233696: Privacy and Security Challenges in Edge-Deployed Generative AI: Threats, Vulnerabilities, and Defense Mechanisms
Mohammed Houache; Djallel Eddine Boubiche; Abdellah Chehri; Larbi Guezouli
1571233835: Distributed Federated Generative-AI Field Learning for Synthetic Sensing in 6G Vehicular Networks
Bhagwan Das; Nawaz Ali; Fiza Siyal; Sandeep Pirbhulal; Ali Hassan Sodhro; Magnus Johnsson
1571236915: Edge-Optimized ExMobileViT: A Lightweight Vision Transformer for Multi-Class Alzheimer's MRI Classification
Fareesa Khan; Audrey Huong; Ali Hassan Sod; Wan Mahani Hafizah Wan Mahmud; Izhar Hussain; Salman Ahmed; Ghulam Hussain; Waheed Ali Laghari
WS01-S2:
Time: 16:00 – 17:30
Room: MR308-309
Chair: Abdellah Chehri
Presentations:
1571239624: TopoEdge: Topology-Grounded Agentic Framework for Edge Networking Code Generation and Repair
Haomin Qi; Bohan Liu; Zihan Dai; Yunkai Gao
1571254377: Transformer Based on Kolmogorov-Arnold Networks for Fine-Grained Network Throughput Prediction
Fan Jiang; JiKe Cai; Xuewei Zhang; Lei Liu; Chaowei Wang
1571254381: Task-Oriented Adaptive Semantic Communication with Multi-Modal Fusion for UAV Object Tracking
Jiaqi Wang; Siyuan Zheng; Yiming Liu