From: Deep Learning-Driven Data Curation and Model Interpretation for Smart Manufacturing
Technique | Advantage | Disadvantage |
---|---|---|
Unrolled optimization | Suited for data with sparse or non-sparse structure; End-to-end learning | Denoising progression not accessible; Intermediate tweak not possible |
Generative prior | Complete denoising progression accessible; Allow intermediate tweak | May not perform well for data with non-sparse structure |