This Dedicated DSI Meet Up will be the fourth of a series on Embedded AI, co-organized by the CAES and DMB groups.
The “Embedded AI” seminar series is designed to bring together students and researchers passionate about embedded artificial intelligence. Our goal is to showcase the latest research and innovative applications in this field while creating a space for networking and collaboration. By attending these talks, you’ll have the opportunity to learn from experts, share your work, and connect with others who are working on similar projects. Whether you’re just starting out or already have experience in embedded AI, this seminar series is a great way to expand your knowledge and be part of a growing community. Join us at our first seminar to learn more about embedded AI's exciting possibilities and how you can contribute to intelligent systems' future.
Join us for an engaging 60-minute BYO-lunch seminar into the captivating world of Embedded AI. During this session, we will cover the following topic:
Abstract
Although Cloud AI remains the primary choice for training and executing Deep Learning (DL) models, its reliance on significant computational resources, dedicated hardware, and associated issues with latency, efficiency, limit its application to a handful of high-end use cases. To achieve cost-effective and pervasive intelligence in emerging AI applications, AI capabilities should be extended from the cloud to the embedded or edge devices, enabling billions of autonomous devices ranging from edge servers to small-scale microcontrollers.
In this talk, I will present the ongoing research at TU Delft's Embedded AI Lab, focusing on our efforts to enhance the energy-, computing-, and data-efficiency of embedded AI systems. I will showcase several systems and applications we have developed over the past few years, such as fingertip air-writing using light, spectrum painting for on-device signal classification, and energy-efficient deep learning for battery-free platforms. Lastly, I will provide a brief overview of the ANT project on Embedded AI Systems, coordinated by TU Delft and funded by the EU MSCA Doctoral Networks program.
Speaker
Dr. Qing Wang received the PhD degree from UC3M and IMDEA Networks Institute, Spain, in 2016. He is currently an assistant professor at the Embedded Systems Group of TU Delft. His research interests include Visible Light Communication and Embedded AI / TinyML for 6G. He is co-founder of TU Delft's Embedded AI Lab. He is the coordinator of ANT, an MSCA-DN project on Embedded AI Systems. He co-founded OpenVLC, an open-source and low-cost platform for VLC research. His research has been published at leading conferences and journals such as MobiCom, MobiSys, SenSys, CoNEXT, ICLR, IMWUT/UbiComp, ToN, JSAC, TMC, TWC, etc. He has received nine paper awards, including six Best Paper Awards (WoWMoM'24, EWSN'23, ICC'23, SenSys'22, Morse'22, COMSNETS'19) and three Best Paper Runner-Ups (EWSN'22, MobiCom'20, CoNEXT'16).
Personal webpage: https://www.st.ewi.tudelft.nl/qing/
