Jinhao Pan

Recommendation models can now be transparently interpretable without sacrificing accuracy, thanks to a novel autoencoder architecture that learns explicit logical rules.