Scaling Laws For Deep Learning Based Image Reconstruction

TL;DR

This paper investigates how increasing training data size affects the performance of deep neural networks in image reconstruction tasks, finding diminishing returns beyond moderate dataset sizes and providing an analytical understanding of this phenomenon.

Contribution

It empirically studies the scaling laws of training set size versus performance in image reconstruction and offers an analytical model explaining the limited gains from larger datasets.

Findings

Performance gains slow significantly at moderate dataset sizes

Training on millions of images would not substantially improve results

Analytical model explains the plateau in performance with increased data

Abstract

Deep neural networks trained end-to-end to map a measurement of a (noisy) image to a clean image perform excellent for a variety of linear inverse problems. Current methods are only trained on a few hundreds or thousands of images as opposed to the millions of examples deep networks are trained on in other domains. In this work, we study whether major performance gains are expected from scaling up the training set size. We consider image denoising, accelerated magnetic resonance imaging, and super-resolution and empirically determine the reconstruction quality as a function of training set size, while simultaneously scaling the network size. For all three tasks we find that an initially steep power-law scaling slows significantly already at moderate training set sizes. Interpolating those scaling laws suggests that even training on millions of images would not significantly improve performance. To understand the expected behavior, we analytically characterize the performance of a linear estimator learned with early stopped gradient descent. The result formalizes the intuition that once the error induced by learning the signal model is small relative to the error floor, more training examples do not improve performance.