Memory-efficient Learning for Large-scale Computational Imaging -- NeurIPS deep inverse workshop
This work addresses memory constraints in GPU-based learning for real-world computational imaging systems, enabling data-driven design in applications like microscopy and MRI.
The paper tackles the memory limitations of training physics-based networks for large-scale computational imaging by proposing a memory-efficient learning procedure that exploits layer reversibility, demonstrating practicality on super-resolution optical microscopy and multi-channel MRI with improved scalability.
Computational imaging systems jointly design computation and hardware to retrieve information which is not traditionally accessible with standard imaging systems. Recently, critical aspects such as experimental design and image priors are optimized through deep neural networks formed by the unrolled iterations of classical physics-based reconstructions (termed physics-based networks). However, for real-world large-scale systems, computing gradients via backpropagation restricts learning due to memory limitations of graphical processing units. In this work, we propose a memory-efficient learning procedure that exploits the reversibility of the network's layers to enable data-driven design for large-scale computational imaging. We demonstrate our methods practicality on two large-scale systems: super-resolution optical microscopy and multi-channel magnetic resonance imaging.