Fig. 3From: Trade-off between conservation of biological variation and batch effect removal in deep generative modeling for single-cell transcriptomicsPareto MTL versus scalarization. In Panels A and C, we show the 12 Pareto candidates produced when MINE and MMD, respectively, are used to measure the effect in the TM-MARROW dataset. In the corresponding Panels B and D, we show the non-dominated points only. When batch effect is measured using MINE, Pareto MTL produces a more diverse set of trade-offs, while scalarization tends to produce trade-offs at the extreme regions. When MMD is the batch effect indication, Pareto MTL seems to perform similarly to scalarizationBack to article page