NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:8774
Title:DeepWave: A Recurrent Neural-Network for Real-Time Acoustic Imaging

The paper proposes a RNN LISTA architecture to the problem of real-time acoustic imaging and a novel parametrization leading to a number of parameters that grow linearly wrt resolution instead of quadratically, as well as a novel initialization scheme. Some Experiments comparing this approach to the state-of-art in the field validate the proposed model. While the paper may of of narrow interest for the ML community, it presents some interesting contributions.