View analytic
Thursday, March 16 • 4:00pm - 5:30pm
Poster: Deep Learning Approaches to Structured Signal Recovery

Sign up or log in to save this to your schedule and see who's attending!

Feedback form is now closed.
The promise of compressive sensing (CS) has been offset by two significant challenges. First,
real-world data is not exactly sparse in a fixed basis. Second, current high-performance recovery
algorithms are slow to converge, which limits CS to either non-real-time applications or scenarios
where massive back-end computing is available. We attack both of these challenges head-on by developing new signal recovery frameworks that learn the inverse transformation from observation vectors to original signals using deep learning techniques. When trained on a set of representative signals, these frameworks learn both a representation for the signals (addressing challenge one) and an inverse map approximating a greedy or convex recovery algorithm addressing challenge two). According to simulation results, our deep network frameworks closely approximate the solution produced by state-of-the-art CS recovery algorithms yet are thousands of times faster
in run time. The trade-off for the ultrafast run time is a computationally intensive, off-line training
procedure typical to deep networks. However, the training needs to be completed only once, which
makes the approach attractive for a host of sparse recovery applications. These applications are but not limited to magnetic resonance imaging (MRI), computed tomography (CT), coded-aperture imaging, and seismic data collection in oil and gas industry.


Thursday March 16, 2017 4:00pm - 5:30pm
Exhibit Hall BRC