Fast Dual-Regularized Autoencoder for Sparse Biological Data
This work addresses sparse data challenges in biological applications like drug discovery, but it is incremental as it builds on an existing linear model.
The authors tackled the problem of relationship inference from sparse biological data by extending a linear model to a shallow autoencoder for dual neighborhood-regularized matrix completion, achieving speed and accuracy advantages over state-of-the-art methods in predicting drug-target interactions and drug-disease associations.
Relationship inference from sparse data is an important task with applications ranging from product recommendation to drug discovery. A recently proposed linear model for sparse matrix completion has demonstrated surprising advantage in speed and accuracy over more sophisticated recommender systems algorithms. Here we extend the linear model to develop a shallow autoencoder for the dual neighborhood-regularized matrix completion problem. We demonstrate the speed and accuracy advantage of our approach over the existing state-of-the-art in predicting drug-target interactions and drug-disease associations.