Susanna Thon, an assistant professor in the Department of Electrical and Computer Engineering, has been selected by the National Science…More
Trac D. Tran Delivers Inaugural Professorial Lecture
As part of the Don P. Giddens Inaugural Professorial Lecture series, Trac D. Tran delivered the lecture entitled “When Less Is More…” Processing signals in the sparsified domain is much faster, simpler, and more robust than in the original domain, and Professor Tran has chronicled the quest for a deeper understanding of this extremely powerful concept and its role in numerous classical signal and information processing applications.
Abstract: In 2020, the universe of global digital information is projected to reach more than 34.6 zettabytes, three times the projected storage capacity. More than 90 percent of this massive data is expected to be unstructured, with a very large portion contributed by smartphones and wearable sensors. By 2020, it is projected that every human being will contribute 1 terabyte of sensor data to this collective information glut. The data will come from technology that includes GPS devices, accelerometers, gyroscopes, microphones, cameras, and various biomedical
One powerful concept that proposes ways to handle these critical data-deluge problems is sparse representation. In other words, most signals from these sensors can be approximately represented by only a few significant components, which carry the most relevant information. Such compact representation not only provides better signal compression for bandwidth/storage efficiency, but since it focuses on the most intrinsic property of the data, it also leads to faster processing algorithms and more effective signal separation for detection, classification, and recognition. Sparse signal representation allows us to capture the hidden simplified structure often present in the data jungle, thus providing necessary compression and minimization of the harmful effects of noise in practical settings. This lecture chronicles the never-ending quest for a deeper understanding of this extremely powerful concept and its role in numerous practical signal and information processing applications.