trainCurveFactors {MOFA} | R Documentation |
The MOFA model starts with an initial number of factors and inactive factors
can be dropped during training if they explain small amounts of variation
(as defined in getDefaultModelOptions
).
This allows the model to automatically infer the dimensionality of the latent space.
The corresponding hyperparameters are defined in prepareMOFA
.
All training statistics, including the number of active factors, can be fetch from the
TrainStats slot of MOFAmodel
.
trainCurveFactors(object)
object |
a |
plot of number of active factors during training
# Example on the CLL data filepath <- system.file("extdata", "CLL_model.hdf5", package = "MOFAdata") MOFA_CLL <- loadModel(filepath) trainCurveFactors(MOFA_CLL) # Example on the scMT data filepath <- system.file("extdata", "scMT_model.hdf5", package = "MOFAdata") MOFA_scMT <- loadModel(filepath) trainCurveFactors(MOFA_scMT)