pipette
ENEnglish

Equivariant Experience Replay for Continual Learning in Imaging Inverse Problems

Guixian Xu, Jin Zhu, Jinglai Li, Junqi Tang

PreprintAfirmaciones fuertes, leer con cuidadoUso en el mundo real

En palabras de los autores

In this work, we propose Equivariant Experience Replay, a new paradigm for achieving continual learning of deep networks for solving ill-posed imaging inverse problems. Inspired by recent advances on unsupervised learning methods such as equivariant imaging/splitting, we find the exploitation of group-symmetry structure is particularly relevant and effective in continual learning setting for ill-posed inversion tasks. We therefore propose this new experience replay scheme, which is tailored for such inverse problems. Our numerical study demonstrates the superiority of our method compared to existing continual learning methods for inverse problems in medical imaging and image processing.

Resultado principalEl resumen no menciona limitaciones.

Apareció: viernes, 23 de octubre. arXiv. Preprint, todavía sin revisión por pares.

Comentario de los autores: 16 pages