Skip to main content

Table 3 Hyperparameters for Pareto MTL

From: Trade-off between conservation of biological variation and batch effect removal in deep generative modeling for single-cell transcriptomics

 

pre-epochs

pre-lr

pre-adv-epochs

pre-adv-lr

epochs

lr

adv-epochs

adv-lr

MMD bandwidths

TM-MARROW

Pareto MTL with MINE

200

\(1e{-}3\)

400

\(5e{-}5\)

150

\(1e{-}3\)

1

\(5e{-}5\)

NA

Pareto MTL with MMD

200

\(1e{-}3\)

NA

NA

250

\(1e{-}3\)

NA

NA

1,2,5,8,10

MACAQUE-RETINA

Pareto MTL with MINE

150

\(1e{-}3\)

400

\(5e{-}5\)

150

\(1e{-}3\)

1

\(5e{-}5\)

NA

Pareto MTL with MMD

150

\(1e{-}3\)

NA

NA

250

\(1e{-}3\)

NA

NA

1,2,5,8,10

  1. The abbreviation “pre” means pre-training, “adv” means MINE adversarial training, “lr” means learning rate