Fig. 1: The architecture of AutoComplete. | Nature Genetics

Fig. 1: The architecture of AutoComplete.

From: Deep learning-based phenotype imputation on population-scale biobank data increases genetic discoveries

Fig. 1

AutoComplete defines a feed-forward encoder-decoder architecture h trained using copy-masking, a procedure that simulates realistic missingness patterns that the model uses to impute missing values. AutoComplete minimizes the loss function \({\mathscr{L}}\) that is defined over the observed and masked values. AutoComplete supports the imputation of continuous and binary features.

Back to article page