Fig. 16
From: Joint embedding–classifier learning for interpretable collaborative filtering
Validation metrics for JELI with different similarity thresholds \(\tau\) on a “second-order” synthetic data set across several iterations (random seeds for splitting between training and testing data sets)