Fig. 4
From: Biologically relevant transfer learning improves transcription factor binding prediction
Transfer learning performs well for TFs with small training datasets. Performance of HNF4A models trained with (blue boxes) and without (red boxes) transfer learning on 5000 (A), 1000 (B), 500 (C), 250 (D), 125 (E), and 50 (F), bound and unbound regions. Each model was trained five times with different random initializations to ensure the robustness of the results. AUCPR, area under the precision-recall curve; TF, transcription factor; TL, transfer learning