Welcome to BPDL’s documentation!¶
Contents¶
- Binary Pattern Dictionary Learning
- bpdl package
- experiments package
- Examples
- Binary Pattern Dictionary Learning
- Generic dictionary learning
- Indipendent Component Analyses
- Multi-subject dictionary learning & CanICA
- Non-negative matrix factorization
- Sparse PCA
- Spectral Clustering
- Synthetic datasets
- Training sigmoid funstion for gene activations
- Sample registration of an image to patterns - Deamons
- All results on Synthetic datasets - Binary images
- BPDL results on Synthetic datasets - binary images
- All results on Synthetic datasets - Prob. images
- Results on Synthetic datasets - Prob. images
Indices and tables¶
BPDL - Binary pattern Dictionary Learning¶
The package contain Binary pattern Dictionary Learning (BPDL) which is image processing toolfor unsupervised pattern extraction and atlas estimation. Moreover the project/repositorycontains comparisons with State-of-the-Art decomposition methods applied to image domain.The package also includes useful tools for dataset handling and around real microscopy images.
Main features¶
implementation of BPDL package
using fuzzy segmentation as inputs
deformation via daemon registration
experimental setting & synthetic dataset
comparison with NMF, SPCS, ICA, CanICA, MSDL, etc.
visualisations and notebook samples
References¶
Borovec J., Kybic J. (2016) Binary Pattern Dictionary Learning for Gene Expression Representationin Drosophila Imaginal Discs. In: Computer Vision - ACCV 2016 Workshops. Lecture Notes in ComputerScience, vol 10117, Springer. DOI: 10.1007/978-3-319-54427-4_40.