Keynotes

Archan Misra Singapore Management University

Abstract Several embodiments of smart computing applications require real-time, on-device perception and response in immersive field environments. Optimizing the execution of heavyweight DNN models for such perception tasks on resource-constrained edge devices remains a formidable challenge. Using two exemplar applications, vehicular traffic tracking and human-robot interactive task execution, I shall describe two powerful and innovative approaches to reduce the complexity, latency and energy of such edge-based DNN inference execution. The first involves the notion of criticality-aware spatiotemporal prioritization of selected regions from multiple video sensor streams, allowing an edge device to achieve 5-8x increase in DNN processing throughput with practically no loss in object detection and tracking accuracy. The second involves the use of "soft-skipping", a dynamic DNN selective execution technique to support low-latency multi-modal human instruction comprehension (combining visual, verbal and gestural input) for natural human-robot interaction. I shall conclude with an overview of open challenges and research directions in the broad area of interactive human-machine task execution.

Bio
Archan Misra is Vice Provost (Research) and Professor of Computer Science at Singapore Management University (SMU). Archan has provided leadership to a number of multi-million dollar, large-scale research initiatives at SMU, including the LiveLabs, LARC and CASA research centers, which collectively developed innovative mobile/wearable computing technologies for urban computing applications. He is a current recipient of the prestigious Investigator grant (from Singapore's National Research Foundation) for sustainable human-machine interaction intelligence. His current research interests lie in ultra-low energy execution of machine perception algorithms on IoT and edge devices to support advances in natural human-machine interaction. Over a 25+ year research career spanning both academics and industry (at IBM Research and Bellcore), Archan has published on, and practically deployed, technologies spanning wireless networking, mobile & wearable sensing and urban mobility analytics. An ACM Distinguished Member, Archan holds a Ph.D. from the University of Maryland at College Park, and chaired the IEEE Computer Society's Technical Committee on Computer Communications (TCCC) from 2005-2007.In 2017, Dr. Ploetz joined the School of Interactive Computing at the Georgia Institute of Technology in Atlanta, USA where he works as an Associate Professor of Computing. Prior to this he was an academic at the School of Computing Science at Newcastle University in Newcastle upon Tyne, UK, where he was a Reader (Assoc. Prof.) for "Computational Behaviour Analysis" affiliated with Open Lab, Newcastle's interdisciplinary research centre for cross-disciplinary research in digital technologies.

Philip V. Orlik Mitsubishi Electric Research Laboratories

Bio Prior to joining MERL in 2000, Phil worked as a simulation engineer for the MITRE Corporation. His current research interests include wireless communications and networking, signal processing for communication systems, queuing theory, and analytical modeling.