Title: Machine Learning Algorithm and Hardware Co-Design Towards Green and Ubiquitous Artificial Intelligence on Both Edge and Cloud

Date: Tuesday, July 30th, 2024

Time: 2:00 pm - 3:30 pm ET

Location: Virtual via Zoom (https://gatech.zoom.us/j/3249649977?pwd=SWgxdHh1VWdaSCtPWnQvdno5ZGFIUT09)

Join our Cloud HD Video Meeting

Zoom is the leader in modern enterprise video communications, with an easy, reliable cloud platform for video and audio conferencing, chat, and webinars across mobile, desktop, and room systems. Zoom Rooms is the original software-based conference room solution used around the world in board, conference, huddle, and training rooms, as well as executive offices and classrooms. Founded in 2011, Zoom helps businesses and organizations bring their teams together in a frictionless environment to get more done. Zoom is a publicly traded company headquartered in San Jose, CA.

gatech.zoom.us



PhD Student:

Haoran You, School of Computer Science, Georgia Institute of Technology

Committee Members:

Dr. Yingyan (Celine) Lin (Advisor) – School of Computer Science, Georgia Institute of Technology

Dr. Calton Pu – School of Computer Science, Georgia Institute of Technology

Dr. Tong Geng – Department of Electrical and Computer Engineering and Computer Science, University of Rochester

Dr. Tushar Krishna – School of Electrical and Computer Engineering, Georgia Institute of Technology

Abstract:

Machine learning (ML) has achieved significant breakthroughs in various applications, such as augmented or virtual reality (AR/AR) and artificial intelligence-generated content (AIGC). However, a substantial research gap exists between the powerful yet large-scale ML models and the limited resources available on edge and cloud computing devices. This gap poses challenges not only for the feasibility of training these models in cloud data centers but also for their deployment on more resource-constrained edge devices, such as smartphones, AR/VR devices, and Internet of Things (IoT) sensors.

To address these challenges, this proposal outlines a co-design strategy that integrates algorithm development with hardware acceleration, fostering a synergy aimed at producing efficient ML systems. For efficient edge computing, the integration of algorithm and hardware co-design and the development of hardware-aware algorithms are explored. For efficient cloud computing, the focus is on designing scalable and efficient ML models for GPUs. Additionally, efficient training and inference techniques are developed to facilitate ML model training and deployment across both edge and cloud platforms. This research proposal aims to develop interrelated thrusts designed to synergize efficient ML models, training and inference algorithms, and hardware accelerators. The objective is to create a unified and efficient ML ecosystem tailored for both edge and cloud platforms. This holistic strategy is crucial for optimizing computational resources and managing data flow across edge and cloud environments.