Embedded Intelligence Towards Smarter and Healthier Environments


Embedded Intelligence Towards Smarter and Healthier Environments

Stephen Xia, Ph.D. Candidate, Department of Electrical Engineering, Columbia University

3:30 p.m.–4:30 p.m., March 17, 2022   |   101 DeBartolo Hall

We have seen a remarkable growth in smart devices and artificial intelligence in all aspects of our  lives. Despite this ever-growing amount of AI around us, our environments are still far from truly  intelligent. At the touch of a button, we have access to powerful AI that can easily outperform any  human in complex tasks and games, like Chess and Go, yet our environments still cannot alert us  to dangerous approaching vehicles, nor prepare a simple meal on a busy day, something all of us  do regularly and intuitively. In this talk, I will present two lines of work that bridge the gap between  AI and truly intelligent environments. 

Stephen Xia

First, I will introduce my work on embedding acoustic intelligence into wearables we commonly  carry, such as headphones and helmets, to create safer cities. These low-cost and long-lasting  wearables leverage novel architectures that utilize a combination of physics-based models and  machine learning techniques to alert pedestrians and construction workers of dangers from  oncoming vehicles. These wearables ultimately act as a second pair of ears, creating a sphere of  safety around us. 

Second, I will present my work on generalized intelligent systems that allow us to embed  intelligence more easily into all aspects of our lives or adapt to the situation at hand. I will first  discuss my work on a generalized selective audio filtering architecture, for resource-constrained  systems, that allows us to embed audio intelligence in diverse applications and platforms. This  architecture dynamically leverages the physics of audio and a wide range of data-driven machine  learning models to allow engineers and developers to enhance and suppress custom sounds in their  own specific applications. Finally, I will introduce my work on a home-centric artificial  intelligence that adapts to the physical environment, its occupants, and available sensors, actuators,  and devices to dynamically learn and carry out a wide range of services and tasks without the need  for humans to manually specify how to carry out each task. This generalized and adaptable  intelligence is a concrete step towards realizing truly intelligent and autonomous environments. 

Stephen Xia is a Ph.D. candidate in the Department of Electrical Engineering at Columbia  University, advised by Dr. Xiaofan (Fred) Jiang. Stephen received his B.S. in Electrical  Engineering from Rice University in 2016. His research lies at the intersection between systems,  embedded machine learning, and signal processing, spanning areas in mobile and embedded  systems, Internet-of-Things, cyber-physical systems, artificial intelligence, and smart health. His  work takes a joint physics-based and data-driven approach to realize truly intelligent and  autonomous environments by embedding and dynamically utilizing compute, sensing, actuation,  storage, and networking resources all around us. Stephen’s research has been highlighted by many  popular media outlets, including Mashable, Fast Company, and Gizmodo, and has received various  distinctions, including Best Demo Awards at ACM SenSys 2021, ACM/IEEE IPSN 2020,  ACM/IEEE IoTDI 2018, and the Best Presentation Award at IEEE VNC 2018.