The maturation of energy harvesting technology and ultra-low-power microcontrollers has led to the advent of tiny, batteryless devices, capable of sustainable, maintenance free operation. Due to remarkable progress in embedded deep learning, these devices can now run deep neural networks (DNN) locally, transitioning intelligence decision making from the cloud to beyond the edge. However, as harvestable ambient energy is weak and unstable, battery-less devices operate intermittently, experiencing frequent power failures, thus presenting unique research challenges. In this talk, Dr. Mendis (Rosh) will first provide an overall introduction to intermittent systems and then present the fundamental challenges with enabling deep learning on battery-less devices. Next, Rosh will highlight some of his work to address these challenges, specifically introducing unique runtime software concepts, to efficiently execute DNN inference on these devices, and novel design automation frameworks that are able to design and customize DNNs for deployment on intermittent systems. Lastly, Rosh will briefly outline his upcoming research direction, which focuses on tackling sustainability concerns associated with deep learning on tiny devices.

Dr. Hashan Roshantha (Rosh) Mendis is a Postdoctoral Scholar (appointed by AS) at the Research Center for Information Technology Innovation (CITI), Academia Sinica, Taiwan. He earned his MSc. and Engineering Doctorate (EngD.) degrees from the Department of Electronics and Department of Computer Science at the University of York in the United Kingdom in 2011 and 2017, respectively. Since 2018, he has been working in CITI's Embedded and Mobile Computing Lab. Prior to that, he gained several years of industry experience, working with renowned companies like Dell Technologies Inc. in Taiwan and innovative SMEs like Yotta Media Labs and PlayJam Games in the UK. His research interests broadly include embedded systems, with a focus on realizing Artificial Intelligence of Things (AIoT).