Virtual Seminar by Jun Luo

Title: Algorithmic Sensing in the Age of Artificial Intelligence

Time and Date: Apr. 8, 2022, at 9:00 am ET

Presenter:  Professor  Jun Luo, Nanyang Technological University, Singapore

Venue: (Password: 4Zn7xZ)

Abstract: Promoted by the increasing compactness of embedded sensors and tremendous progress in artificial intelligence, algorithmic sensing is gaining momentum as it has the potential to push the limit of physical sensors. However, realizing the promises of algorithmic sensing can be highly non-trivial, as specific machine/deep learning techniques have to be fine-tuned to suit diversified sensing modalities, as opposed to direct feature extraction and classification largely ignoring the distinctive physics of respective sensors. In this talk, we take Radio Frequency (RF) sensing as a representative example. We first explain the basic idea of RF-based algorithmic sensing, then we present a unified framework to prepare RF sensing data for deep analytics. Building upon this framework, we further explain the peculiarity of RF sensing data (against, e.g., conventional image data) and present our approaches in designing deep interpreted RF sensing, leveraging our latest developments on a (software-defined sensing) platform, along with corresponding applications and products as instances.

Bio: Dr. LUO Jun is a tenured professor in NTU. He has been working on wireless sensing, deep learning, and computing system integration for more than two decades and has established a solid foundation on all these topics. Building upon this foundation, his research team has made contributions on mobile/ pervasive computing and smart sensing technologies, by leading several national research projects and corporate labs (including MoE Tier2, BMW, SAP, and CSIJRI, with more than S$10 million funding), aiming to transfer the research outcomes to practical applications.
In the meantime, Dr. Luo and his team have kept publishing at relevant venues such as IEEE and ACM top conferences and journals, including MobiCom, CVPR, SenSys, INFOCOM, UbiComp, ToN, and TMC. With only 130+ publications, he has earned over 8000 Google Scholar citations, with the top-cited paper bearing more than 1,200 of them. In the past few years, Dr. Luo has put most his efforts on algorithmic sensing aiming to, on one hand, push sensing capabilities far beyond conventional sensor design using novel machine learning techniques, and on the other hand, repurpose existing sensors to achieve functionalities beyond their original intentions. More detailed information can be found at: