Abstract:
The next breakthrough in artificial intelligence will be driven not only by advances in models and accelerators, but by innovations in communication networks. As AI systems scale from thousands to millions of distributed accelerators, communication is rapidly replacing computation as the primary bottleneck.
AI systems massive data movement demand networking architectures that provide unprecedented levels of bandwidth, ultra-low latency, resilience, security, and energy efficiency. The network is becoming an integral part of the AI system. Today's communication infrastructures face serious challenges in scalability, reliability, security, energy consumption, operational complexity, maintenance and management as we continue to use retrofitted techniques and protocols designed for communications and technologies that existed several decades ago. These limitations threaten to constrain the future of large-scale AI training, distributed AI inference, and emerging autonomous AI ecosystems. Addressing these challenges requires rethinking networking from first principles and developing architectures and techniques suited to AI-native traffic models and workloads with a capability to evolve with trends and demands.
This presentation explores the evolving AI ecosystem and the multidisciplinary research challenges spanning compute, memory, storage, networking, scheduling, resilience, software, cybersecurity, and sustainable infrastructure. Emphasis is placed on networking as a foundational enabler of future AI systems and on the emerging research challenges in AI-native network architectures, scalable routing and multipath forwarding, high-performance transport protocols, network telemetry and AI-driven control, scheduler-network co-design, resilient communication and trustworthy AI networks.
Recognizing this critical need, the IEEE Rochester Joint AESS/Communications Society Chapter has established a Special Interest Group (SIG) on Networking for AI Infrastructure to bring together researchers, industry leaders, and standards organizations to advance the next generation of AI communication systems. The SIG aims to foster collaboration, accelerate innovation, and participate in standards to shape the future of communication infrastructures not limited to the next generation of AI.
Speaker(s): Prof. Dr. Nirmala Shenoy,
Agenda:
Pizza
Introduction of the Speaker
Talk
Q&A
Room: Grand/Large Conference Room, Bldg: Nashua Public Library, 2 Court Street, NASHUA, New Hampshire, United States, 03060, Virtual: https://events.vtools.ieee.org/m/571705
