Data Analytics for Complex Systems: From Circuit Design to Brain Analysis
Tuesday, May 3, 2016
11:45 am - 1:00 pm
Gross Hall 330
Xin Li, PhD Associate Professor, ECE Department, Carnegie Mellon University
Data analytics is an important area that has been continuously growing during the past several decades. This seminar will present several novel statistical algorithms and methodologies for two application domains: circuit and brain. First, a statistical technique, referred to as virtual probe (VP), is proposed to efficiently measure, characterize and monitor spatially-correlated variations for nanoscale integrated circuits. VP exploits the recent breakthroughs in compressive sensing to accurately predict spatial variations from an exceptionally small set of measurement data. Second, a robust regression technique, called robust signal space separation (rSSS), is adopted to model the magnetic field generated by human brain based on quasi-static Maxwell equation. rSSS is then used to remove the noise/artifact and, hence, improve the signal-to-noise ratio for magnetoencephalography. Finally, a number of on-going projects for data analytics will be briefly discussed, covering several interdisciplinary fields such as biomedical engineering, autonomous driving, early child education, business intelligence, advanced manufacturing, etc.