A novel machine learning based approach for iPS progenitor cell identification
In all sub-figures, X axis indicates the start frame of the best time windows and the corresponding window length (13 frames) is indicated in the inlet. (a) 5-fold cross-validation precisions over 10 runs. (b) the standard deviation of the average precision of the neighborhood time windows in Fig 6D. (c) the standard deviation of the average precision of the distant windows in Fig 6E. (d) the average precision of seven neighborhood time windows calculated over 10 holdout validation runs. (e) the average precision over 10 independent tests for six best time windows on their corresponding distant windows.