BioMoR Autoencoder and Embeddings

This vignette illustrates the intended role of autoencoders within BioMoR.

Currently, the autoencoder functionality is implemented as a lightweight stub that returns placeholder embeddings. This allows users to experiment with the high-level pipeline without requiring heavy deep-learning dependencies.

# Example: training a stub autoencoder and obtaining embeddings

data(iris)
feature_cols <- c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width")

ae <- train_autoencoder(iris, feature_cols = feature_cols, epochs = 10)
emb <- get_embeddings(ae, iris, feature_cols = feature_cols)

str(emb)