Precision Learning: Towards Use of Known Operators in Neural Networks
This work addresses the challenge of integrating prior knowledge into neural networks for applications in physics and signal processing, offering an incremental improvement by leveraging known transforms.
The paper tackles the problem of incorporating known operators or transforms into neural networks to improve performance, demonstrating that this approach reduces prediction errors, as shown by an increase in SSIM values from 0.54 to 0.88 in an X-ray material decomposition example.
In this paper, we consider the use of prior knowledge within neural networks. In particular, we investigate the effect of a known transform within the mapping from input data space to the output domain. We demonstrate that use of known transforms is able to change maximal error bounds. In order to explore the effect further, we consider the problem of X-ray material decomposition as an example to incorporate additional prior knowledge. We demonstrate that inclusion of a non-linear function known from the physical properties of the system is able to reduce prediction errors therewith improving prediction quality from SSIM values of 0.54 to 0.88. This approach is applicable to a wide set of applications in physics and signal processing that provide prior knowledge on such transforms. Also maximal error estimation and network understanding could be facilitated within the context of precision learning.