Closed tulerpetontidae closed 2 years ago
Regularisation with:
[x] Just grid-search the best sigma parameter
[x] Learn Laplace sigma parameter
[x] Learn Laplace sigma for each gene
[x] Learn Laplace sigma as a product of gene and cell type specific parameters
Transformations:
[x] z-score after log1p(x / sum )
[x] x / std
[x] pseudo z-score ?
The aim is to compare on train/validation split each of these cases
Regularisation with:
[x] Just grid-search the best sigma parameter
[x] Learn Laplace sigma parameter
[x] Learn Laplace sigma for each gene
[x] Learn Laplace sigma as a product of gene and cell type specific parameters
Transformations:
[x] z-score after log1p(x / sum )
[x] x / std
[x] pseudo z-score ?
The aim is to compare on train/validation split each of these cases