University of Twente Student Theses
Multi-scale convolutional neural networks using partial differential equations
Jutte, A.M.P. (2019) Multi-scale convolutional neural networks using partial differential equations.
|
PDF
1MB |
| Abstract: | Convolutional neural networks (CNN) are known for their superior performance over other classical methods in numerous applications. A downside of conventional CNNs is that they are trained on a specific image size and are only meant to be used for this image size. Training a network on high resolution images is expensive. Downscaling high resolution images such that they can be used for low resolution networks results in loss on information. Multi-scale CNNs are capable of performing on different image sizes. In this paper a multi-scale method is presented based on partial differential equations. Using partial differential equations, a continuous representation can be created for neural networks of specific structures. The results presented in this paper are focused on denoising applications. However, the method presented in this paper could have applications in for example the removal of artefacts from CT scans and the removal of timestamps from images. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 31 mathematics |
| Programme: | Applied Mathematics BSc (56965) |
| Link to this item: | https://purl.utwente.nl/essays/79102 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page