“Time Series Data Analysis via Deep Learning for Mobile and Embedded Computing”
JeongGil Ko is an associate professor in the School of Integrated Technology, College of Engineering at Yonsei University. He received his Bachelors in Engineering (B.Eng.) degree in computer science and engineering from Korea University in 2007 and received his Doctor of Philosophy (Ph.D.) degree in Computer Science from the Johns Hopkins University in 2012. At Johns Hopkins, JeongGil Ko was a member of the Hopkins interNetworking Research Group (HiNRG) led by Dr. Andreas Terzis. In 2010, he was at the Stanford Information Networking Group (SING) with Dr. Philip Levis at Stanford University as a visiting researcher. Before joining Yonsei University in September 2019, JeongGil Ko was an assistant professor at the Department of Software and Computer Engineering at Ajou University from September 2015 to August 2019, and was a senior researcher at the Electronics and Telecommunications Research Institute (ETRI) from June 2012 to August 2015. He is a recipient of the Abel Wolman Fellowship awarded by the Whiting School of Engineering at the Johns Hopkins University in 2007 and a senior member of the IEEE since 2017. His research interests are in the general area of developing embedded and mobile computing systems with ambient intelligence.
This talk will discuss two mobile/embedded systems that involve the analysis of time series data (one numerical and the other image) using deep learning models. The first part of the talk presents HeartQuake, a low cost, accurate, non-intrusive, geophone-based sensing system for extracting accurate electrocardiogram (ECG) patterns using heartbeat vibrations that penetrate through a bed mattress. In HeartQuake, cardiac activity-originated vibration patterns are captured on a geophone and sent to a server, where the data is filtered to remove the sensor’s internal noise and passed on to a bidirectional long short term memory (Bi-LSTM) deep learning model for ECG waveform estimation. Our extensive experimental results with baseline dataset collected from 21 study participants and a longitudinal dataset from 15 study participants suggest that HeartQuake, even when using a general non-personalized model, can detect all five ECG peaks (e.g., P, Q, R, S, T) with an average error of as low as 13 msec when participants are stationary on the bed. Furthermore, clinically used ECG metrics such as RR interval and QRS segment width can be estimated with errors as low as 3 msec and 10 msec, respectively. When additional noise factors are present (e.g., external vibration and various sleeping habits), the estimation error increases, but can be mitigated by using a personalized model. Finally, a qualitative study with 11 physicians on the clinically perceived quality of HeartQuake-generated ECG signals suggests that HeartQuake can effectively serve as a screening tool for detecting and diagnosing abnormal cardiovascular conditions.
ZOOM : https://zoom.us/j/8978217407?pwd=d1pOZmF1OWlseEdZRVBpV3VuSkl3dz09